It seems to be unexpectedly passing on some bots probably because it requires asserts to fail, but doesn't say that. But we already have a patch in review to make it not xfail so I'd rather just focus on getting it passing rather than trying to figure out an unexpected pass.
llvm-svn: 348661
We were still using the rounded down offset and alignment even though
they aren't handled because you can't trivially bitcast the loaded
value.
llvm-svn: 348658
To make X86CondBrFoldingPass can be run with --run-pass option, this can test one wrong assertion on analyzeCompare function for SUB32ri when its operand is not imm
Patch by Jianping Chen
Differential Revision: https://reviews.llvm.org/D55412
llvm-svn: 348620
As discussed in the post-commit thread of r347917, this
transform is fighting with an existing transform causing
an infinite loop or out-of-memory, so this is effectively
reverting r347917 and its follow-up r348195 while we
investigate the bug.
llvm-svn: 348604
This change attempts to shrink scalar AND, OR and XOR instructions which take an immediate that isn't inlineable.
It performs:
AND s0, s0, ~(1 << n) -> BITSET0 s0, n
OR s0, s0, (1 << n) -> BITSET1 s0, n
AND s0, s1, x -> ANDN2 s0, s1, ~x
OR s0, s1, x -> ORN2 s0, s1, ~x
XOR s0, s1, x -> XNOR s0, s1, ~x
In particular, this catches setting and clearing the sign bit for fabs (and x, 0x7ffffffff -> bitset0 x, 31 and or x, 0x80000000 -> bitset1 x, 31).
llvm-svn: 348601
When we had dynamic call frames (i.e. sp adjustment around each call) we
were including that adjustment into offsets calculated based on r6, even
though it's only sp that changes. This led to incorrect stack slot
accesses.
llvm-svn: 348591
Adds fatal errors for any target that does not support the Tiny or Kernel
codemodels by rejigging the getEffectiveCodeModel calls.
Differential Revision: https://reviews.llvm.org/D50141
llvm-svn: 348585
Fix assert about using an undefined physical register in machine instruction verify pass.
The reason is that register flag undef is missing when doing transformation from If Conversion Pass.
```
Bad machine code: Using an undefined physical register
- function: func_65
- basic block: %bb.0 entry (0x10024740738)
- instruction: BCLR killed $cr5lt, implicit $lr8, implicit $rm, implicit undef $x3
- operand 0: killed $cr5lt
LLVM ERROR: Found 1 machine code errors.
```
There are also other existing testcases with same issue. So I add -verify-machineinstrs option to open verifying.
Differential Revision: https://reviews.llvm.org/D55408
llvm-svn: 348566
If this is not a valid way to assign an SDLoc, then we get this
wrong all over SDAG.
I don't know enough about the SDAG to explain this. IIUC, theoretically,
debug info is not supposed to affect codegen. But here it has clearly
affected 3 different targets, and the x86 change is an actual improvement.
llvm-svn: 348552
This was probably organized as it was because bswap is a unary op.
But that's where the similarity to the other opcodes ends. We should
not limit this transform to scalars, and we should not try it if
either input has other uses. This is another step towards trying to
clean this whole function up to prevent it from causing infinite loops
and memory explosions.
Earlier commits in this series:
rL348501
rL348508
rL348518
llvm-svn: 348534
This patch introduces a new DAGCombiner rule to simplify concat_vectors nodes:
concat_vectors( bitcast (scalar_to_vector %A), UNDEF)
--> bitcast (scalar_to_vector %A)
This patch only partially addresses PR39257. In particular, it is enough to fix
one of the two problematic cases mentioned in PR39257. However, it is not enough
to fix the original test case posted by Craig; that particular case would
probably require a more complicated approach (and knowledge about used bits).
Before this patch, we used to generate the following code for function PR39257
(-mtriple=x86_64 , -mattr=+avx):
vmovsd (%rdi), %xmm0 # xmm0 = mem[0],zero
vxorps %xmm1, %xmm1, %xmm1
vblendps $3, %xmm0, %xmm1, %xmm0 # xmm0 = xmm0[0,1],xmm1[2,3]
vmovaps %ymm0, (%rsi)
vzeroupper
retq
Now we generate this:
vmovsd (%rdi), %xmm0 # xmm0 = mem[0],zero
vmovaps %ymm0, (%rsi)
vzeroupper
retq
As a side note: that VZEROUPPER is completely redundant...
I guess the vzeroupper insertion pass doesn't realize that the definition of
%xmm0 from vmovsd is already zeroing the upper half of %ymm0. Note that on
%-mcpu=btver2, we don't get that vzeroupper because pass vzeroupper insertion
%pass is disabled.
Differential Revision: https://reviews.llvm.org/D55274
llvm-svn: 348522
The PPC test with 2 extra uses seems clearly better by avoiding this transform.
With 1 extra use, we also prevent an extra register move (although that might
be an RA problem). The general rule should be to only make a change here if
it is always profitable. The x86 diffs are all neutral.
llvm-svn: 348518
This reverts commit r348203 and reapplies D55085 with an additional
GCOV bugfix to make the change NFC for relative file paths in .gcno files.
Thanks to Ilya Biryukov for additional testing!
Original commit message:
Update Diagnostic handling for changes in CFE.
The clang frontend no longer emits the current working directory for
DIFiles containing an absolute path in the filename: and will move the
common prefix between current working directory and the file into the
directory: component.
https://reviews.llvm.org/D55085
llvm-svn: 348512
The AVX512 diffs are neutral, but the bswap test shows a clear overreach in
hoistLogicOpWithSameOpcodeHands(). If we don't check for other uses, we can
increase the instruction count.
This could also fight with transforms trying to go in the opposite direction
and possibly blow up/infinite loop. This might be enough to solve the bug
noted here:
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20181203/608593.html
I did not add the hasOneUse() checks to all opcodes because I see a perf
regression for at least one opcode. We may decide that's irrelevant in the
face of potential compiler crashing, but I'll see if I can salvage that first.
llvm-svn: 348508
The code emitting AND-subtrees used to check whether any of the operands
was an OR in order to figure out if the result needs to be negated.
However the OR could be hidden in further subtrees and not immediately
visible.
Change the code so that canEmitConjunction() determines whether the
result of the generated subtree needs to be negated. Cleanup emission
logic to use this. I also changed the code a bit to make all negation
decisions early before we actually emit the subtrees.
This fixes http://llvm.org/PR39550
Differential Revision: https://reviews.llvm.org/D54137
llvm-svn: 348444
These opcodes are intended to subsume some of the capability of G_MERGE_VALUES,
as it was too powerful and thus complex to add deal with throughout the GISel
pipeline.
G_BUILD_VECTOR creates a vector value from a sequence of uniformly typed
scalar values. G_BUILD_VECTOR_TRUNC is a special opcode for handling scalar
operands which are larger than the destination vector element type, and
therefore does an implicit truncate.
G_CONCAT_VECTOR creates a vector by concatenating smaller, uniformly typed,
vectors together.
These will be used in a subsequent commit. This commit just adds the initial
infrastructure.
Differential Revision: https://reviews.llvm.org/D53594
llvm-svn: 348430
Mostly NFC, only change is the order of outlined function names.
Loop over the outlined functions instead of walking the candidate list.
This is a bit easier to understand. It's far more natural to create a function,
then replace all of its occurrences with calls than the other way around.
The functions outlined after this do not change, but their names will be
decided by their benefit. E.g, OUTLINED_FUNCTION_0 will now always be the
most beneficial function, rather than the first one seen.
This makes it easier to enforce an ordering on the outlined functions. So,
this also adds a test to make sure that the ordering works as expected.
llvm-svn: 348414
Because we're potentially peeking through a bitcast in this transform,
we need to use overall bitwidths rather than number of elements to
determine when it's safe to proceed.
Should fix:
https://bugs.llvm.org/show_bug.cgi?id=39893
llvm-svn: 348383
Whenever we effectively take the address of a basic block we need to
manually update that basic block to reflect that fact or later passes
such as tail duplication and tail merging can break the invariants of
the code. =/ Sadly, there doesn't appear to be any good way of
automating this or even writing a reasonable assert to catch it early.
The change seems trivially and obviously correct, but sadly the only
really good test case I have is 1000s of basic blocks. I've tried
directly writing a test case that happens to make tail duplication do
something that crashes later on, but this appears to require an
*amazingly* complex set of conditions that I've not yet reproduced.
The change is technically covered by the tests because we mark the
blocks as having their address taken, but that doesn't really count as
properly testing the functionality.
llvm-svn: 348374
fidelity checking of RIP-based references to basic blocks and other
labels.
These labels are super important for SLH tests so we should keep them
readable in the test cases.
llvm-svn: 348373
This is an initial patch to add a minimum level of support for funnel shifts to the SelectionDAG and to begin wiring it up to the X86 SHLD/SHRD instructions.
Some partial legalization code has been added to handle the case for 'SlowSHLD' where we want to expand instead and I've added a few DAG combines so we don't get regressions from the existing DAG builder expansion code.
Differential Revision: https://reviews.llvm.org/D54698
llvm-svn: 348353
Functions annotated with `__fastcall` or `__attribute__((__fastcall__))`
or `__attribute__((__swiftcall__))` may contain SEH handlers even on
Win64. This matches the behaviour of cl which allows for
`__try`/`__except` inside a `__fastcall` function. This was detected
while trying to self-host clang on Windows ARM64.
llvm-svn: 348337
It looks like MCRegAliasIterator can visit the same physical register twice. When this happens in this code in LICM we end up setting the PhysRegDef and then later in the same loop visit the register again. Now we see that PhysRegDef is set from the earlier iteration so now set PhysRegClobber.
This patch splits the loop so we have one that uses the previous value of PhysRegDef to update PhysRegClobber and second loop that updates PhysRegDef.
The X86 atomic test is an improvement. I had to add sideeffect to the two shrink wrapping tests to prevent hoisting from occurring. I'm not sure about the AMDGPU tests. It looks like the branch instruction changed at end the of the loops. And in the branch-relaxation test I think there is now "and vcc, exec, -1" instruction that wasn't there before.
Differential Revision: https://reviews.llvm.org/D55102
llvm-svn: 348330
We previously disabled this in r323371 because of a bug where we selected an
extending load, but didn't delete the old G_LOAD, resulting in two loads being
generated for volatile loads.
Since we now have dedicated G_SEXTLOAD/G_ZEXTLOAD operations, and that the
tablegen patterns should no longer be able to select (ext(load x)) patterns, it
should be safe to re-enable it.
The old test case should still work as expected.
llvm-svn: 348320
Currently if you use -{start,stop}-{before,after}, it picks
the first instance with the matching pass name. If you run
the same pass multiple times, there's no way to distinguish them.
Allow specifying a run index wih ,N to specify which you mean.
llvm-svn: 348285
This reverts commit r348203.
Reason: this produces absolute paths in .gcno files, breaking us
internally as we rely on them being consistent with the filenames passed
in the command line.
Also reverts r348157 and r348155 to account for revert of r348154 in
clang repository.
llvm-svn: 348279
PR17686 demonstrates that for some targets FP exceptions can fire in cases where the FP_TO_UINT is expanded using a FP_TO_SINT instruction.
The existing code converts both the inrange and outofrange cases using FP_TO_SINT and then selects the result, this patch changes this for 'strict' cases to pre-select the FP_TO_SINT input and the offset adjustment.
The X87 cases don't need the strict flag but generates much nicer code with it....
Differential Revision: https://reviews.llvm.org/D53794
llvm-svn: 348251
Add support for ISD::*_EXTEND and ISD::*_EXTEND_VECTOR_INREG opcodes.
The extra broadcast in trunc-subvector.ll will be fixed in an upcoming patch.
llvm-svn: 348246
The comment was misplaced, and the code didn't do what the comment indicated,
namely ignoring the varargs portion when computing the local stack size of a
funclet in emitEpilogue. This results in incorrect offset computations within
funclets that are contained in vararg functions.
Differential Revision: https://reviews.llvm.org/D55096
llvm-svn: 348222
This moves the stack check logic into a lambda within getOutliningCandidateInfo.
This allows us to be less conservative with stack checks. Whether or not a
stack instruction is safe to outline is dependent on the frame variant and call
variant of the outlined function; only in cases where we modify the stack can
these be unsafe.
So, if we move that logic later, when we're looking at an individual candidate,
we can make better decisions here.
This gives some code size savings as a result.
llvm-svn: 348220
This is the smallest vector enhancement I could find to D54640.
Here, we're allowing narrowing to only legal vector ops because we'll see
regressions without that. All of the test diffs are wins from what I can tell.
With AVX/AVX512, we can shrink ymm/zmm ops to xmm.
x86 vector multiplies are the problem case that we're avoiding due to the
patchwork ISA, and it's not clear to me if we can dance around those
regressions using TLI hooks or if we need preliminary patches to plug those
holes.
Differential Revision: https://reviews.llvm.org/D55126
llvm-svn: 348195
If it's a bigger code size win to drop candidates that require stack fixups
than to demote every candidate to that variant, the outliner should do that.
This happens if the number of bytes taken by calls to functions that don't
require fixups, plus the number of bytes that'd be left is less than the
number of bytes that it'd take to emit a save + restore for all candidates.
Also add tests for each possible new behaviour.
- machine-outliner-compatible-candidates shows that when we have candidates
that don't use the stack, we can use the default call variant along with the
no save/regsave variant.
- machine-outliner-all-stack shows that when it's better to fix up the stack,
we still will demote all candidates to that case
- machine-outliner-drop-stack shows that we can discard candidates that
require stack fixups when it would be beneficial to do so.
llvm-svn: 348168
Summary:
We need to unpackl and unpackh the operands to use two vXi16 multiplies. Previously it looks like the low unpack would get constant folded at least in the 128-bit case after shuffle lowering turned the unpackl into ZERO_EXTEND_VECTOR_INREG and X86 custom DAG combined it. The same doesn't happen for the high half. So we'd load a constant and then shuffle it. But the low half would just be loaded and used by the multiply directly.
After this patch we now end up with a constant pool entry for the low and high unpacks separately with no shuffle operations.
This is a step towards removing custom constant folding for ZERO_EXTEND_VECTOR_INREG/SIGN_EXTEND_VECTOR_INREG in the X86 backend.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D55165
llvm-svn: 348159
Summary:
Under -x86-experimental-vector-widening-legalization, fp_to_uint/fp_to_sint with a smaller than 128 bit vector type results are custom type legalized by promoting the result to a 128 bit vector by promoting the elements, inserting an assertzext/assertsext, then truncating back to original type. The truncate will be further legalizdd to a pack shuffle. In the case of a v8i8 result type, we'll end up with a v8i16 fp_to_sint. This will need to be further legalized during vector op legalization by promoting to v8i32 and then truncating again. Under avx2 this produces good code with two pack instructions, but Under avx512 this will result in a truncate instruction and a packuswb instruction. But we should be able to get away with a single truncate instruction.
The other option is to promote all the way to vXi32 result type during the first type legalization. But in some experimentation that seemed to require more work to produce good code for other configurations.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54836
llvm-svn: 348158
The clang frontend no longer emits the current working directory for
DIFiles containing an absolute path in the filename: and will move the
common prefix between current working directory and the file into the
directory: component.
https://reviews.llvm.org/D55085
llvm-svn: 348155
The introduction of S_{ADD|SUB}_U64_PSEUDO instructions which are decomposed
into VOP3 instruction pairs for S_ADD_U64_PSEUDO:
V_ADD_I32_e64
V_ADDC_U32_e64
and for S_SUB_U64_PSEUDO
V_SUB_I32_e64
V_SUBB_U32_e64
preclude the use of SDWA to encode a constant.
SDWA: Sub-Dword addressing is supported on VOP1 and VOP2 instructions,
but not on VOP3 instructions.
We desire to fold the bit-and operand into the instruction encoding
for the V_ADD_I32 instruction. This requires that we transform the
VOP3 into a VOP2 form of the instruction (_e32).
%19:vgpr_32 = V_AND_B32_e32 255,
killed %16:vgpr_32, implicit $exec
%47:vgpr_32, %49:sreg_64_xexec = V_ADD_I32_e64
%26.sub0:vreg_64, %19:vgpr_32, implicit $exec
%48:vgpr_32, dead %50:sreg_64_xexec = V_ADDC_U32_e64
%26.sub1:vreg_64, %54:vgpr_32, killed %49:sreg_64_xexec, implicit $exec
which then allows the SDWA encoding and becomes
%47:vgpr_32 = V_ADD_I32_sdwa
0, %26.sub0:vreg_64, 0, killed %16:vgpr_32, 0, 6, 0, 6, 0,
implicit-def $vcc, implicit $exec
%48:vgpr_32 = V_ADDC_U32_e32
0, %26.sub1:vreg_64, implicit-def $vcc, implicit $vcc, implicit $exec
Differential Revision: https://reviews.llvm.org/D54882
llvm-svn: 348132
This has two positive effects. First, using a custom node prevents
recombination leading to an infinite loop since the output DAG is notionally a
little more complex than the input one. Using a flag-setting instruction also
allows the subtraction to be folded with the related comparison more easily.
https://reviews.llvm.org/D53190
llvm-svn: 348122
Summary:
There are 4 instructions which have Inconsistent ImmMustBeMultipleOf in the
function PPCInstrInfo::instrHasImmForm, they are LFS, LFD, STFS, STFD.
These four instructions should set the ImmMustBeMultipleOf to 1 instead of 4.
Reviewed By: steven.zhang
Differential Revision: https://reviews.llvm.org/D54738
llvm-svn: 348109
This makes the SDAG behavior consistent with the way we do this in IR.
It's possible that we were getting the wrong answer before. For example,
'xor undef, undef --> 0' but 'xor undef, C' --> undef.
But the most practical improvement is likely as shown in the tests here -
for FP, we were overconstraining undef lanes to NaN, and that can prevent
vector simplifications/narrowing (see D51553).
llvm-svn: 348090
The generic legalizer will fall back to a stack spill that uses a truncating store. That store will get expanded into a shuffle and non-truncating store on pre-avx512 targets. Once that happens the stack store/load pair will be combined away leaving behind the shuffle and bitcasts. On avx512 targets the truncating store is legal so doesn't get folded away.
By custom legalizing it we can avoid this churn and maybe produce better code.
llvm-svn: 348085
If we know that we'll definitely save LR to a register, there's no reason to
pre-check whether or not a stack instruction is unsafe to fix up.
This makes it so that we check for that condition before mapping instructions.
This allows us to outline more, since we don't pessimise as many instructions.
Also update some tests, since we outline more.
llvm-svn: 348081
Summary: With sse4.1 we use two zero_extend_vector_inreg and a pshufd to expand the v16i8 input into two v8i16 vectors for the multiply. That's 3 shuffles to extend one operand. The other operand is usually constant as this is mostly used by division by constant optimization. Pre sse4.1 we use a punpckhbw and a punpcklbw with a zero vector. That's two shuffles and an xor and a copy due to tied register constraints. That seems maybe better than the 3 shuffles. With AVX we avoid the copy so that's obviously better.
Reviewers: spatel, RKSimon
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D55138
llvm-svn: 348079
The identity ~(x ^ y) == (~x ^ y) == (x ^ ~y) allows XNOR (XOR/NOT) to turn into NOT/XOR. Handling this case with its own split means we can make the NOT remain in the scalar unit. Previously, we split 64-bit XNOR into two 32-bit XNOR, then lowered. Now, we get three instructions (s_not, v_xor, v_xor) rather than four in the case where either of the sources is a scalar 64-bit.
Add test cases to xnor.ll to attempt XNOR Vx, Sy and XNOR Sx, Vy. Also adding test that uses the opposite identity such that (~x ^ y) on the scalar unit (or vector for gfx906) can generate XNOR. This already worked, but I didn't see a test for it.
Differential: https://reviews.llvm.org/D55071
llvm-svn: 348075
D52935 introduced the ability for SimplifyDemandedBits to call SimplifyDemandedVectorElts through BITCASTs if the demanded bit mask entirely covered the sub element.
This patch relaxes this to demanding an element if we need any bit from it.
Differential Revision: https://reviews.llvm.org/D54761
llvm-svn: 348073
As noted by Eli Friedman <https://reviews.llvm.org/D52977?id=168629#1315291>,
the RV64I shift patterns for SLLW/SRLW/SRAW make some incorrect assumptions.
SRAW assumed that (sext_inreg foo, i32) could only be produced when
sign-extended an i32. However, it can be produced by input such as:
define i64 @tricky_ashr(i64 %a, i64 %b) {
%1 = shl i64 %a, 32
%2 = ashr i64 %1, 32
%3 = ashr i64 %2, %b
ret i64 %3
}
It's important not to select sraw in the above case, because sraw only uses
bits lower 5 bits from the shift, while a shift of 32-63 would be valid.
Similarly, the patterns for srlw assumed (and foo, 0xffffffff) would only be
produced when zero-extending a value that was originally i32 in LLVM IR. This
is obviously incorrect.
This patch removes the SLLW/SRLW/SRAW shift patterns for the time being and
adds test cases that would demonstrate a miscompile if the incorrect patterns
were re-added.
llvm-svn: 348067
Summary:
Moving SMRD to VMEM in SIFixSGPRCopies is rather bad for performance if
the load is really uniform. So select the scalar load intrinsics directly
to either VMEM or SMRD buffer loads based on divergence analysis.
If an offset happens to end up in a VGPR -- either because a floating
point calculation was involved, or due to other remaining deficiencies
in SIFixSGPRCopies -- we use v_readfirstlane.
There is some unrelated churn in tests since we now select MUBUF offsets
in a unified way with non-scalar buffer loads.
Change-Id: I170e6816323beb1348677b358c9d380865cd1a19
Reviewers: arsenm, alex-t, rampitec, tpr
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D53283
llvm-svn: 348050
Summary:
The VirtReg2Value mapping is crucial for getting consistently
reliable divergence information into the SelectionDAG. This
patch fixes a bunch of issues that lead to incorrect divergence
info and introduces tight assertions to ensure we don't regress:
1. VirtReg2Value is generated lazily; there were some cases where
a lookup was performed before all relevant virtual registers were
created, leading to an out-of-sync mapping. Those cases were:
- Complex code to lower formal arguments that generated CopyFromReg
nodes from live-in registers (fixed by never querying the mapping
for live-in registers).
- Code that generates CopyToReg for formal arguments that are used
outside the entry basic block (fixed by never querying the
mapping for Register nodes, which don't need the divergence info
anyway).
2. For complex values that are lowered to a sequence of registers,
all registers must be reflected in the VirtReg2Value mapping.
I am not adding any new tests, since I'm not actually aware of any
bugs that these problems are causing with trunk as-is. However,
I recently added a test case (in r346423) which fails when D53283 is
applied without this change. Also, the new assertions should provide
most of the effective test coverage.
There is one test change in sdwa-peephole.ll. The underlying issue
is that since the divergence info is now correct, the DAGISel will
select V_OR_B32 directly instead of S_OR_B32. This leads to an extra
COPY which affects the behavior of MachineLICM in a way that ends up
with the S_MOV_B32 with the constant in a different basic block than
the V_OR_B32, which is presumably what defeats the peephole.
Reviewers: alex-t, arsenm, rampitec
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D54340
llvm-svn: 348049
Instead of treating the outlined functions for these as distinct frames, they
should be combined into one case. Neither allows for stack fixups, and both
generate the same frame. Thus, they ought to be considered one case.
This makes the code far easier to understand, for one thing. It also offers
some small code size improvements. It's fairly rare to see a class of outlined
functions that doesn't fall entirely into one variant (on CTMark anyway). It
does happen from time to time though.
This mostly offers some serious simplification.
Also update the test to show the added functionality.
llvm-svn: 348036
All that you can legitimately do with the CFI for a nounwind function
is get a backtrace, and adjusting the SCS register is not (currently)
required for this purpose.
Differential Revision: https://reviews.llvm.org/D54988
llvm-svn: 348035
This reduces the number of shuffle operations that need to be done. The splitting strategy requires the shuffle unit for the extraction and the extension. With the unpack strategy the unpacks accomplish a splitting and extending in one operation.
llvm-svn: 348019
This does require a constant pool load instead of loading an immediate into a gpr, moving to a k register and masking. But its less instructions and more consistent with previous ISAs. It probably opens up more combine opportunities as one of the test cases demonstrates.
llvm-svn: 348018
Summary:
If a given liveness arg of STATEPOINT is at a fixed frame index
(e.g. a function argument passed on stack), prefer to use this
fixed location even the address is also in a register. If we use
the register it will generate a spill, which is not necessary
since the fixed frame index can be directly recorded in the stack
map.
Patch by Cherry Zhang <cherryyz@google.com>.
Reviewers: thanm, niravd, reames
Reviewed By: reames
Subscribers: cherryyz, reames, anna, arphaman, llvm-commits
Differential Revision: https://reviews.llvm.org/D53889
llvm-svn: 347998
Introduces DPP pseudo instructions and the pass that combines DPP mov with subsequent uses.
Differential revision: https://reviews.llvm.org/D53762
llvm-svn: 347993
For targets where i32 is not a legal type (e.g. 64-bit RISC-V),
LegalizeIntegerTypes must promote the result of ISD::FLT_ROUNDS_.
Differential Revision: https://reviews.llvm.org/D53820
llvm-svn: 347986
For targets where i32 is not a legal type (e.g. 64-bit RISC-V),
LegalizeIntegerTypes must promote the operands of ISD::PREFETCH.
Differential Revision: https://reviews.llvm.org/D53281
llvm-svn: 347980
For targets where i32 is not a legal type (e.g. 64-bit RISC-V),
LegalizeIntegerTypes must promote the operand.
Differential Revision: https://reviews.llvm.org/D53279
llvm-svn: 347978
DAGTypeLegalizer::PromoteSetCCOperands currently prefers to zero-extend
operands when it is able to do so. For some targets this is more expensive
than a sign-extension, which is also a valid choice. Introduce the
isSExtCheaperThanZExt hook and use it in the new SExtOrZExtPromotedInteger
helper. On RISC-V, we prefer sign-extension for FromTy == MVT::i32 and ToTy ==
MVT::i64, as it can be performed using a single instruction.
Differential Revision: https://reviews.llvm.org/D52978
llvm-svn: 347977
As discussed in the RFC
<http://lists.llvm.org/pipermail/llvm-dev/2018-October/126690.html>, 64-bit
RISC-V has i64 as the only legal integer type. This patch introduces patterns
to support codegen of the new instructions
introduced in RV64I: addiw, addiw, subw, sllw, slliw, srlw, srliw, sraw,
sraiw, ld, sd.
Custom selection code is needed for srliw as SimplifyDemandedBits will remove
lower bits from the mask, meaning the obvious pattern won't work:
def : Pat<(sext_inreg (srl (and GPR:$rs1, 0xffffffff), uimm5:$shamt), i32),
(SRLIW GPR:$rs1, uimm5:$shamt)>;
This is sufficient to compile and execute all of the GCC torture suite for
RV64I other than those files using frameaddr or returnaddr intrinsics
(LegalizeDAG doesn't know how to promote the operands - a future patch
addresses this).
When promoting i32 sltu/sltiu operands, it would be more efficient to use
sign-extension rather than zero-extension for RV64. A future patch adds a hook
to allow this.
Differential Revision: https://reviews.llvm.org/D52977
llvm-svn: 347973
Don't expand SDIV with an immediate that is a power of 2 if we optimise for
minimum code size. For example:
sdiv %1, i32 4
gets expanded to a sequence of 3 instructions, but this is suboptimal for
minimum code size so instead we just generate a MOV and a SDIV if integer
division is supported.
Differential Revision: https://reviews.llvm.org/D54546
llvm-svn: 347965
The motivating case for this is shown in:
https://bugs.llvm.org/show_bug.cgi?id=32023
and the corresponding rot16.ll regression tests.
Because x86 scalar shift amounts are i8 values, we can end up with trunc-binop-trunc
sequences that don't get folded in IR.
As the TODO comments suggest, there will be regressions if we extend this (for x86,
we mostly seem to be missing LEA opportunities, but there are likely vector folds
missing too). I think those should be considered existing bugs because this is the
same transform that we do as an IR canonicalization in instcombine. We just need
more tests to make those visible independent of this patch.
Differential Revision: https://reviews.llvm.org/D54640
llvm-svn: 347917
Utilise a similar ('late') lowering strategy to D47882. The changes to
AtomicExpandPass allow this strategy to be utilised by other targets which
implement shouldExpandAtomicCmpXchgInIR.
All cmpxchg are lowered as 'strong' currently and failure ordering is ignored.
This is conservative but correct.
Differential Revision: https://reviews.llvm.org/D48131
llvm-svn: 347914
Also revert fix r347876
One of the buildbots was reporting a failure in some relevant tests that I can't
repro or explain at present, so reverting until I can isolate.
llvm-svn: 347911
It makes more sense to order FI-based memops in descending order when
the stack goes down. This allows offsets to stay "consecutive" and allow
easier pattern matching.
llvm-svn: 347906
I believe we should be legalizing these with the rest of vector binary operations. If any custom lowering is required for these nodes, this will give the DAG combine between LegalizeVectorOps and LegalizeDAG to run on the custom code before constant build_vectors are lowered in LegalizeDAG.
I've moved MULHU/MULHS handling in AArch64 from Lowering to isel. Moving the lowering earlier caused build_vector+extract_subvector simplifications to kick in which made the generated code worse.
Differential Revision: https://reviews.llvm.org/D54276
llvm-svn: 347902
This is another patch for -x86-experimental-vector-widening. This pre widens narrow division by constants so that we can get pass the legal type check in the generic DAG combiner. Otherwise we end up scalarizing.
I've restricted this to splats for now because it was easy to just call DAG.getConstant. Not sure what we should do for non-splat? Increase the element size?Widen the constant vector by padding with 1?
Differential Revision: https://reviews.llvm.org/D54919
llvm-svn: 347898
Summary:
Replace `aext([asz]ext x)` with `aext/sext/zext x` in order to
reduce the number of instructions generated to clean up some
legalization artifacts.
Reviewers: aditya_nandakumar, dsanders, aemerson, bogner
Reviewed By: aemerson
Subscribers: rovka, kristof.beyls, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D54174
llvm-svn: 347893
This patch adds support for S_ANDN2, S_ORN2 32-bit and 64-bit instructions and adds splits to move them to the vector unit (for which there is no equivalent instruction). It modifies the way that the more complex scalar instructions are lowered to vector instructions by first breaking them down to sequences of simpler scalar instructions which are then lowered through the existing code paths. The pattern for S_XNOR has also been updated to apply inversion to one input rather than the output of the XOR as the result is equivalent and may allow leaving the NOT instruction on the scalar unit.
A new tests for NAND, NOR, ANDN2 and ORN2 have been added, and existing tests now hit the new instructions (and have been modified accordingly).
Differential: https://reviews.llvm.org/D54714
llvm-svn: 347877
TFE and LWE support requires extra result registers that are written in the
event of a failure in order to detect that failure case.
The specific use-case that initiated these changes is sparse texture support.
This means that if image intrinsics are used with either option turned on, the
programmer must ensure that the return type can contain all of the expected
results. This can result in redundant registers since the vector size must be a
power-of-2.
This change takes roughly 6 parts:
1. Modify the instruction defs in tablegen to add new instruction variants that
can accomodate the extra return values.
2. Updates to lowerImage in SIISelLowering.cpp to accomodate setting TFE or LWE
(where the bulk of the work for these instruction types is now done)
3. Extra verification code to catch cases where intrinsics have been used but
insufficient return registers are used.
4. Modification to the adjustWritemask optimisation to account for TFE/LWE being
enabled (requires extra registers to be maintained for error return value).
5. An extra pass to zero initialize the error value return - this is because if
the error does not occur, the register is not written and thus must be zeroed
before use. Also added a new (on by default) option to ensure ALL return values
are zero-initialized that is required for sparse texture support.
6. Disable the inst_combine optimization in the presence of tfe/lwe (later TODO
for this to re-enable and handle correctly).
There's an additional fix now to avoid a dmask=0
For an image intrinsic with tfe where all result channels except tfe
were unused, I was getting an image instruction with dmask=0 and only a
single vgpr result for tfe. That is incorrect because the hardware
assumes there is at least one vgpr result, plus the one for tfe.
Fixed by forcing dmask to 1, which gives the desired two vgpr result
with tfe in the second one.
The TFE or LWE result is returned from the intrinsics using an aggregate
type. Look in the test code provided to see how this works, but in essence IR
code to invoke the intrinsic looks as follows:
%v = call {<4 x float>,i32} @llvm.amdgcn.image.load.1d.v4f32i32.i32(i32 15,
i32 %s, <8 x i32> %rsrc, i32 1, i32 0)
%v.vec = extractvalue {<4 x float>, i32} %v, 0
%v.err = extractvalue {<4 x float>, i32} %v, 1
Differential revision: https://reviews.llvm.org/D48826
Change-Id: If222bc03642e76cf98059a6bef5d5bffeda38dda
llvm-svn: 347871
It causes asserts building BoringSSL. See https://crbug.com/91009#c3 for
repro.
This also reverts the follow-ups:
Revert r347724 "Do not insert prefetches with unsupported memory operands."
Revert r347606 "[X86] Add dependency from X86 to ProfileData after rL347596"
Revert r347607 "Add new passes to X86 pipeline tests"
llvm-svn: 347864
* Tell the StackProtector pass to generate the epilogue instrumentation
when GlobalISel is enabled because GISel currently does not implement
the same deferred epilogue insertion as SelectionDAG.
* Update StackProtector::InsertStackProtectors() to find a stack guard
slot by searching for the llvm.stackprotector intrinsic when the
prologue was not created by StackProtector itself but the pass still
needs to generate the epilogue instrumentation. This fixes a problem
when the pass would abort because the stack guard AllocInst pointer
was null when generating the epilogue -- test
CodeGen/AArch64/GlobalISel/arm64-irtranslator-stackprotect.ll.
Differential Revision: https://reviews.llvm.org/D54518
llvm-svn: 347862
Summary:
MachineLoopInfo cannot be relied on for correctness, because it cannot
properly recognize loops in irreducible control flow which can be
introduced by late machine basic block optimization passes. See the new
test case for the reduced form of an example that occurred in practice.
Use a simple fixpoint iteration instead.
In order to facilitate this change, refactor WaitcntBrackets so that it
only tracks pending events and registers, rather than also maintaining
state that is relevant for the high-level algorithm. Various accessor
methods can be removed or made private as a consequence.
Affects (in radv):
- dEQP-VK.glsl.loops.special.{for,while}_uniform_iterations.select_iteration_count_{fragment,vertex}
Fixes: r345719 ("AMDGPU: Rewrite SILowerI1Copies to always stay on SALU")
Reviewers: msearles, rampitec, scott.linder, kanarayan
Subscribers: arsenm, kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits, hakzsam
Differential Revision: https://reviews.llvm.org/D54231
llvm-svn: 347853
Summary:
Reduce the statefulness of the algorithm in two ways:
1. More clearly split generateWaitcntInstBefore into two phases: the
first one which determines the required wait, if any, without changing
the ScoreBrackets, and the second one which actually inserts the wait
and updates the brackets.
2. Communicate pre-existing s_waitcnt instructions using an argument to
generateWaitcntInstBefore instead of through the ScoreBrackets.
To simplify these changes, a Waitcnt structure is introduced which carries
the counts of an s_waitcnt instruction in decoded form.
There are some functional changes:
1. The FIXME for the VCCZ bug workaround was implemented: we only wait for
SMEM instructions as required instead of waiting on all counters.
2. We now properly track pre-existing waitcnt's in all cases, which leads
to less conservative waitcnts being emitted in some cases.
s_load_dword ...
s_waitcnt lgkmcnt(0) <-- pre-existing wait count
ds_read_b32 v0, ...
ds_read_b32 v1, ...
s_waitcnt lgkmcnt(0) <-- this is too conservative
use(v0)
more code
use(v1)
This increases code size a bit, but the reduced latency should still be a
win in basically all cases. The worst code size regressions in my shader-db
are:
WORST REGRESSIONS - Code Size
Before After Delta Percentage
1724 1736 12 0.70 % shaders/private/f1-2015/1334.shader_test [0]
2276 2284 8 0.35 % shaders/private/f1-2015/1306.shader_test [0]
4632 4640 8 0.17 % shaders/private/ue4_elemental/62.shader_test [0]
2376 2384 8 0.34 % shaders/private/f1-2015/1308.shader_test [0]
3284 3292 8 0.24 % shaders/private/talos_principle/1955.shader_test [0]
Reviewers: msearles, rampitec, scott.linder, kanarayan
Subscribers: arsenm, kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits, hakzsam
Differential Revision: https://reviews.llvm.org/D54226
llvm-svn: 347848
Summary:
A signed comparison of i1 values produces the opposite result to an unsigned one if the condition code
includes less-than or greater-than. This is so because 1 is the most negative signed i1 number and the
most positive unsigned i1 number. The CR-logical operations used for such comparisons are non-commutative
so for signed comparisons vs. unsigned ones, the input operands just need to be swapped.
Reviewed By: steven.zhang
Differential Revision: https://reviews.llvm.org/D54825
llvm-svn: 347831
This failed to select (which might be a separate bug) in
X86ISelDAGToDAG because we try to create a select node
that can be simplified away after rL347227.
This change avoids the problem by simplifying the SHRUNKBLEND
node sooner. In the test case, we manage to realize that the
true/false values of the select (SHRUNKBLEND) are the same thing,
so it simplifies away completely.
llvm-svn: 347818
Expansion of SIGN_EXTEND_INREG can create a VSRAI instruction. If there is already a VSRAI after it, we should combine them into a larger VSRAI
Differential Revision: https://reviews.llvm.org/D54959
llvm-svn: 347784
Before this patch, the following stores in `merge_fail` would fail to be
merged, while they would get merged in `merge_ok`:
```
void use(unsigned long long *);
void merge_fail(unsigned key, unsigned index)
{
unsigned long long args[8];
args[0] = key;
args[1] = index;
use(args);
}
void merge_ok(unsigned long long *dst, unsigned a, unsigned b)
{
dst[0] = a;
dst[1] = b;
}
```
The reason is that `getMemOpBaseImmOfs` would return false for FI base
operands.
This adds support for this.
Differential Revision: https://reviews.llvm.org/D54847
llvm-svn: 347747
This is skylake-avx512 with the addition of avx512vnni ISA.
Patch by Jianping Chen
Differential Revision: https://reviews.llvm.org/D54785
llvm-svn: 347681
If we fold the bitcast into the store we'll end up creating a truncating store to vXi1 that will get scalarized. Instead allow the bitcast to be turned into a movmsk.
We probably need to do something if the store itself is a vXi1 type, but I'll leave that til a testcase appears.
llvm-svn: 347632
Currently a store combine will absorb the bitcast before our combine that turns bitcasts into movmsk gets a chance to run. This results in a store being created with a vXi1 type. Type legalization then promotes the input type and makes this a truncating store. Then we badly scalarize this store.
Currently we avoid this on v8i1->i8 bitcasts due to an incompletely qualified(per the original intention) check in isLoadBitCastBeneficial. An easy fix is to disable this for all vXi1->iX bitcasts on pre-avx512 targets. We'll still generate terrible code if the IR explicitly contains a store of vXi1 without a bitcast. We could probably solve that by just turning all stores of vXi1 into (store (iX (bitcast))) as an early DAG combine.
llvm-svn: 347631
Summary:
Support for profile-driven cache prefetching (X86)
This change is part of a larger system, consisting of a cache prefetches recommender, create_llvm_prof (https://github.com/google/autofdo), and LLVM.
A proof of concept recommender is DynamoRIO's cache miss analyzer. It processes memory access traces obtained from a running binary and identifies patterns in cache misses. Based on them, it produces a csv file with recommendations. The expectation is that, by leveraging such recommendations, we can reduce the amount of clock cycles spent waiting for data from memory. A microbenchmark based on the DynamoRIO analyzer is available as a proof of concept: https://goo.gl/6TM2Xp.
The recommender makes prefetch recommendations in terms of:
* the binary offset of an instruction with a memory operand;
* a delta;
* and a type (nta, t0, t1, t2)
meaning: a prefetch of that type should be inserted right before the instrution at that binary offset, and the prefetch should be for an address delta away from the memory address the instruction will access.
For example:
0x400ab2,64,nta
and assuming the instruction at 0x400ab2 is:
movzbl (%rbx,%rdx,1),%edx
means that the recommender determined it would be beneficial for a prefetchnta instruction to be inserted right before this instruction, as such:
prefetchnta 0x40(%rbx,%rdx,1)
movzbl (%rbx, %rdx, 1), %edx
The workflow for prefetch cache instrumentation is as follows (the proof of concept script details these steps as well):
1. build binary, making sure -gmlt -fdebug-info-for-profiling is passed. The latter option will enable the X86DiscriminateMemOps pass, which ensures instructions with memory operands are uniquely identifiable (this causes ~2% size increase in total binary size due to the additional debug information).
2. collect memory traces, run analysis to obtain recommendations (see above-referenced DynamoRIO demo as a proof of concept).
3. use create_llvm_prof to convert recommendations to reference insertion locations in terms of debug info locations.
4. rebuild binary, using the exact same set of arguments used initially, to which -mllvm -prefetch-hints-file=<file> needs to be added, using the afdo file obtained at step 3.
Note that if sample profiling feedback-driven optimization is also desired, that happens before step 1 above. In this case, the sample profile afdo file that was used to produce the binary at step 1 must also be included in step 4.
The data needed by the compiler in order to identify prefetch insertion points is very similar to what is needed for sample profiles. For this reason, and given that the overall approach (memory tracing-based cache recommendation mechanisms) is under active development, we use the afdo format as a syntax for capturing this information. We avoid confusing semantics with sample profile afdo data by feeding the two types of information to the compiler through separate files and compiler flags. Should the approach prove successful, we can investigate improvements to this encoding mechanism.
Reviewers: davidxl, wmi, craig.topper
Reviewed By: davidxl, wmi, craig.topper
Subscribers: davide, danielcdh, mgorny, aprantl, eraman, JDevlieghere, llvm-commits
Differential Revision: https://reviews.llvm.org/D54052
llvm-svn: 347596
SplitVecOp_TruncateHelper tries to promote the result type while splitting FP_TO_SINT/UINT. It then concatenates the result and introduces a truncate to the original result type. But it does this without inserting the AssertZExt/AssertSExt that the regular result type promotion would insert. Nor does it turn FP_TO_UINT into FP_TO_SINT the way normal result type promotion for these operations does. This is bad on X86 which doesn't support FP_TO_SINT until AVX512.
This patch disables the use of SplitVecOp_TruncateHelper for these operations and just lets normal promotion handle it. I've tweaked a couple things in X86ISelLowering to avoid a few obvious regressions there. I believe all the changes on X86 are improvements. The other targets look neutral.
Differential Revision: https://reviews.llvm.org/D54906
llvm-svn: 347593
We might find a target specific node that needs to be unwrapped after we look through an add/or. Otherwise we get inconsistent results if one pointer is just X86WrapperRIP and the other is (add X86WrapperRIP, C)
Differential Revision: https://reviews.llvm.org/D54818
llvm-svn: 347591
Summary:
STATEPOINT records its args' locations on stack relative to SP.
If the SP is changed, take that into account.
This patch authored by Cherry Zhang <cherryyz@google.com>.
Reviewers: thanm, reames
Reviewed By: reames
Subscribers: reames, llvm-commits
Differential Revision: https://reviews.llvm.org/D53603
llvm-svn: 347569
We have these 2 "isDesirable" promotion hooks (I'm not sure why we need both of them, but that's
independent of this patch), and we can adjust them to promote "mul i8 X, C" to i32. Then, all of
our existing LEA and other multiply expansion magic happens as it would for i32 ops.
Some of the test diffs show that we could end up with an actual 32-bit mul instruction here
because we choose not to expand to simpler ops. That instruction could be slower depending on the
subtarget. On the plus side, this means we don't need a separate instruction to load the constant
operand and possibly an extra instruction to move the result. If we need to tune mul i32 further,
we could add a later transform that tries to shrink it back to i8 based on subtarget timing.
I did not bother to duplicate all of the 32-bit test file RUNs and target settings that exist to
test whether LEA expansion is cheap or not. The diffs here assume a default target, so that means
LEA is generally cheap.
Differential Revision: https://reviews.llvm.org/D54803
llvm-svn: 347557
We can now select CLZ via the TableGen'erated code, so support G_CTLZ
and G_CTLZ_ZERO_UNDEF throughout the pipeline for types <= s32.
Legalizer:
If the CLZ instruction is available, use it for both G_CTLZ and
G_CTLZ_ZERO_UNDEF. Otherwise, use a libcall for G_CTLZ_ZERO_UNDEF and
lower G_CTLZ in terms of it.
In order to achieve this we need to add support to the LegalizerHelper
for the legalization of G_CTLZ_ZERO_UNDEF for s32 as a libcall (__clzsi2).
We also need to allow lowering of G_CTLZ in terms of G_CTLZ_ZERO_UNDEF
if that is supported as a libcall, as opposed to just if it is Legal or
Custom. Due to a minor refactoring of the helper function in charge of
this, we will also allow the same behaviour for G_CTTZ and G_CTPOP.
This is not going to be a problem in practice since we don't yet have
support for treating G_CTTZ and G_CTPOP as libcalls (not even in
DAGISel).
Reg bank select:
Map G_CTLZ to GPR. G_CTLZ_ZERO_UNDEF should not make it to this point.
Instruction select:
Nothing to do.
llvm-svn: 347545
Both zext and sext are currently allowed during the search for narrow
sequences and sexts operands are later added to the mac candidates.
But operands of muls are also added, without checking whether they're
sext or zext, which means we can generate a signed smlad when we
shouldn't.
Differential Revision: https://reviews.llvm.org/D54790
llvm-svn: 347542
This reverts commits r347532. Forget add the option
-mtriple powerpc64-unknown-linux-gnu. So other platform is error except
for PowerPC.
llvm-svn: 347534
When splitting the v16f32/v8f64 result type, type legalization will try to promote the integer result type before a concat and an explicit truncate. But for the fptoui test case this is particularly bad since fptoui isn't supported on X86 until AVX512. We could use an fptosi since the result range would fit in a signed 32-bit value, but the generic type legalization doesn't do that transformation when splitting. It does do this when promoting.
llvm-svn: 347533
Summary:
There are 4 instructions which have Inconsistent ImmMustBeMultipleOf in the
function PPCInstrInfo::instrHasImmForm, they are LFS, LFD, STFS, STFD.
These four instructions should set the ImmMustBeMultipleOf to 1 instead of 4.
Reviewed By: nemanjai
Differential Revision: https://reviews.llvm.org/D54738
llvm-svn: 347532
This should likely be adjusted to limit this transform
further, but these diffs should be clear wins.
If we have blendv/conditional move, then we should assume
those are cheap ops. The loads become independent of the
compare, so those can be speculated before we need to use
the values in the blend/mov.
llvm-svn: 347526
There are many options here depending on subtarget,
but we are uniformly relying on a transform that was
driven by performance for a 32-bit SSE2 target in 2009.
Note: The same motivation was apparently used to do this
transform for *all* targets, so non-x86 may want to look
at this too.
llvm-svn: 347525
...and use them to avoid creating obviously undef values as
discussed in the post-commit thread for r347478.
The diffs in vector div/rem show that we were missing real
optimizations by creating bogus shift nodes.
llvm-svn: 347502
I'm not sure if this actually preserves the original intent
of this test, but if we leave it as-is, the -1 (oversized)
shift should be folded to undef and allow deleting half
of the output.
llvm-svn: 347501
I am working on making FileCheck stricter (in D54769 and D53710) so that it
issues diagnostics when there's something wrong with tests.
This is a cleanup for dangling prefixes in the ARM codegen tests, e.g.:
--check-prefixes=A,B
where A occurs in the check file, but B doesn't. This can be innocent if A does
all the required checking, but can also be a bug in that test if it results in
the test actually not checking anything (if A for example only checks a common
label). Test CodeGen/ARM/smml.ll is such an example.
Differential Revision: https://reviews.llvm.org/D54842
llvm-svn: 347487
This code takes a truncate, fp_to_int, or int_to_fp with a legal result type and an input type that needs to be split and enlarges the elements in the result type before doing the split. Then inserts a follow up truncate or fp_round after concatenating the two halves back together.
But if the input type of the original op is being split on its way to ultimately being scalarized we're just going to end up building a vector from scalars and then truncating or rounding it in the vector register. Seems kind of silly to enlarge the result element type of the operation only to end up with scalar code and then building a vector with large elements only to make the elements smaller again in the vector register. Seems better to just try to get away producing smaller result types in the scalarized code.
The X86 test case that changes is a pretty contrived test case that exists because of a bug we used to have in our AVG matching code. I think the code is better now, but its not realistic anyway.
llvm-svn: 347482
SplitVecOp_TruncateHelper tries to introduce a multilevel truncate to avoid scalarization. But if splitting the result type would still be a legal type we don't need to do that.
The comment block at the top of the function implied that this was already implemented. I looked back through the history and it doesn't look to have ever been checked.
llvm-svn: 347479
We fail to canonicalize IR this way (prefer 'not' ops to arbitrary 'xor'),
but that would not matter without this patch because DAGCombiner was
reversing that transform. I think we need this transform in the backend
regardless of what happens in IR to catch cases where the shift-xor
is formed late from GEP or other ops.
https://rise4fun.com/Alive/NC1
Name: shl
Pre: (-1 << C2) == C1
%shl = shl i8 %x, C2
%r = xor i8 %shl, C1
=>
%not = xor i8 %x, -1
%r = shl i8 %not, C2
Name: shr
Pre: (-1 u>> C2) == C1
%sh = lshr i8 %x, C2
%r = xor i8 %sh, C1
=>
%not = xor i8 %x, -1
%r = lshr i8 %not, C2
https://bugs.llvm.org/show_bug.cgi?id=39657
llvm-svn: 347478
A consequence of r347274 is that SCALAR_TO_VECTOR can be converted into
BUILD_VECTOR by SimplifyDemandedBits, but LowerBUILD_VECTOR can turn
BUILD_VECTOR into SCALAR_TO_VECTOR so we get an infinite loop.
Fix this by making LowerBUILD_VECTOR not do this transformation for those
vectors that would get transformed back, i.e. BUILD_VECTOR of a single-element
constant vector. Doing that means we get a DUP, which we then need to recognise
in ISel as a copy.
llvm-svn: 347456
r334871 has made it possible for TableGen'erated code to select BFC, but
it has not added a test for it on the ARM side. Add it now to make sure
we don't introduce regressions if we ever change anything about that
rule.
llvm-svn: 347447
GCC does it this way, and we have to be consistent. This includes
stdcall and fastcall functions with suffixes. I confirmed that a
fastcall function named "foo" ends up in ".text$foo", not
".text$@foo@8".
Based on a patch by Andrew Yohn!
Fixes PR39218.
Differential Revision: https://reviews.llvm.org/D54762
llvm-svn: 347431
We have efficient codegen on P9 for lowering bswap that involves moving
the value into a vector reg and moving it back. However, the check under
which we custom lowered it did not adequately reflect the actual requirements.
It required only that the subtarget be an implementation of ISA 3.0 since all
compliant implementations have to provide the vector instructions.
However, the kernel builds have a valid use case for -mno-altivec -mcpu=pwr9
(i.e. don't emit vector code, don't have to save vector regs for context
switch). So we should require the correct features for this lowering.
Fixes https://bugs.llvm.org/show_bug.cgi?id=39334
llvm-svn: 347376
These are AVX2 instructions, but have been incorrectly marked in tablegen for a while. This wasn't a problem until r346784 switched the patterns to use target independent ISD opcodes. This made the patterns visible to fast isel.
Fixes PR39733
llvm-svn: 347375
We can't guarantee that demanded bits passing through the vector shuffle won't cause the AND in front of this to be removed. This would prevent the PACKUS from being matched during shuffle lowering.
Unfortunately, this adds a packuswb to one of the vector-reduce-mul.ll tests since we were removing the shuffle via SimplifyDemandedVectorElts. We appear to have similar issues with vpmovwb on the same test case on other targets.
llvm-svn: 347361
This is another step in vector narrowing - a follow-up to D53784
(and hoping to eventually squash potential regressions seen in
D51553).
The x86 test diffs are wins, but the AArch64 diff is probably not.
That problem already exists independent of this patch (see PR39722), but it
went unnoticed in the previous patch because there were no regression tests
that showed the possibility.
The x86 diff in i64-mem-copy.ll is close. Given the frequency throttling
concerns with using wider vector ops, an extra extract to reduce vector
width is the right trade-off at this level of codegen.
Differential Revision: https://reviews.llvm.org/D54392
llvm-svn: 347356
Previously we emitted to separate shuffles, one for unpcklbw and one for unpcklwd. Instead emit a single shuffle equivalent to both of the original shuffles. Shuffle lowering seems able to handle it. This avoids a bitcast between the two shuffles which seems helpful to DAG combine.
Remove the custom type legalization for v8i8->v8i32. I had put that in to avoid some almost duplicate punpcklbw instructions I was seeing, but this lowering change seems to fix that. It also fixes some duplicate shuffles seen in vector-sext.ll
llvm-svn: 347348
Rather than assuming that `tempRet0` exists in linear memory only assume
the getter/setter functions exist. This avoids conflicting with
binaryen which declares a wasm global for this purpose and defines it's
own getter and setter for that.
The other advantage of doing things this way is that it leaving
it up to the linker/finalizer to decide how to actually store this
temporary. As it happens binaryen uses a wasm global which is more
appropriate since it is thread safe.
This also allows us to change the way this is stored in the future
(memory, TLS memory, wasm global) without modifying LLVM.
This is part of a 4 part change:
LLVM: https://reviews.llvm.org/D53240
fastcomp: https://github.com/kripken/emscripten-fastcomp/pull/237
emscripten: https://github.com/kripken/emscripten/pull/7358
binaryen: https://github.com/WebAssembly/binaryen/pull/1709
Differential Revision: https://reviews.llvm.org/D53240
llvm-svn: 347340
This uncovered an off-by-one typo in SimplifyDemandedVectorElts's INSERT_SUBVECTOR handling as its bounds check was bailing on safe indices.
llvm-svn: 347313
When doing some instruction scheduling work, we noticed some missing itineraries.
Before we switch to machine scheduler, those missing itineraries might not have impact to actually scheduling,
because we can still get same latency due to default values.
With machine scheduler, however, itineraries will have impact to scheduling.
eg: NumMicroOps will default to be 0 if there is NO itineraries for specific instruction class.
And most of the instruction class with itineraries will have NumMicroOps default to 1.
This will has impact on the count of RetiredMOps, affects the Pending/Available Queue,
then causing different scheduling or suboptimal scheduling further.
This patch is for STWU/STWUX (IIC_LdStStoreUpd ) for P8.
Since there are already multiple IIC for store update, this patch also merge
IIC_LdStSTDU/IIC_LdStStoreUpd to IIC_LdStSTU
IIC_LdStSTDUX to IIC_LdStSTUX
and we add a new testcase in https://reviews.llvm.org/D54699 to show the difference.
Differential Revision: https://reviews.llvm.org/D54700
llvm-svn: 347311
This patch add a STWU testcase for scheduling check.
Currently P7/P8 which use itineraries are missing IIC_LdStStoreUpd,
We use CHECK-ITIN prefix to check P7/P8, then use default for P9 (and future).
We will fix the missing itineraries of IIC_LdStStoreUpd in following patch,
and update this testcase to show the scheduling difference only there.
Differential Revision: https://reviews.llvm.org/D54699
llvm-svn: 347310
Pull out getPackDemandedElts demanded elts remapping helper from computeKnownBitsForTargetNode and use in computeKnownBits/ComputeNumSignBits.
llvm-svn: 347303
For bitcast nodes from larger element types, add the ability for SimplifyDemandedVectorElts to call SimplifyDemandedBits by merging the elts mask to a bits mask.
I've raised https://bugs.llvm.org/show_bug.cgi?id=39689 to deal with the few places where SimplifyDemandedBits's lack of vector handling is a problem.
Differential Revision: https://reviews.llvm.org/D54679
llvm-svn: 347301
Previously if V2 was unused we ended up using V1 for both inputs as part of the code that follows the new code. By using lowerVectorShuffleWithUNPCK we keep the undef nature of V2 in the output.
As near as I can tell this makes v16i8 behavior consistent with every other VT now.
This does mean that we give the register allocator freedom to fill in random registers now and create false dependencies. But like I said we're already doing that for other types.
llvm-svn: 347296
getZeroVector produces a specifically canonicalized zero vector, but we can just let DAG legalization take care of it.
The test changes are because MULH lowering happens later than it should and this change gave us the opportunity to constant fold away a multiply during a DAG combine before the build_vector got legalized with a bitcast.
llvm-svn: 347290
Turns out that there was no check for a store that truncates down
to a single byte when combining a (store (bswap...)) into a byte-swapping
store. This patch just adds that check.
Fixes https://bugs.llvm.org/show_bug.cgi?id=39478.
llvm-svn: 347288
Summary:
We already support this for scalars, but it was explicitly disabled for vectors. In the updated test cases this allows us to see the upper bits are zero to use less multiply instructions to emulate a 64 bit multiply.
This should help with this ispc issue that a coworker pointed me to https://github.com/ispc/ispc/issues/1362
Reviewers: spatel, efriedma, RKSimon, arsenm
Reviewed By: spatel
Subscribers: wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D54725
llvm-svn: 347287
This can occur when one of the inputs to the multiply is loop invariant. Though my test cases just use two basic blocks with an unconditional jump which we won't merge until after isel in the codegen pipeline.
For scalars, I believe SelectionDAGBuilder can add an AssertZExt to pass knowledge across basic blocks but its explicitly disabled for vectors.
llvm-svn: 347266
This works if DAG combiner is enabled, but without combining
we cannot select scalar_to_vector of <2 x half> and <2 x i16>.
Differential Revision: https://reviews.llvm.org/D54718
llvm-svn: 347259
As discussed on D53794, for float types with ranges smaller than the destination integer type, then we should be able to just use a regular FP_TO_SINT opcode.
I thought we'd need to provide MSA test cases for very small integer types as well (fp16 -> i8 etc.), but it turns out that promotion will kick in so they're unnecessary.
Differential Revision: https://reviews.llvm.org/D54703
llvm-svn: 347251
SSE PSHUFB vector ctlz lowering works at the i4 nibble level. As detailed in PR39703, we were masking the lower nibble off but we only actually use it in the case where the upper nibble is known to be zero, making it safe to remove the mask and save an instruction.
Differential Revision: https://reviews.llvm.org/D54707
llvm-svn: 347242
Previously we split the vectors in half to allow the two halves to be any extended then concatenated the results back together.
This patch instead instead extends the v16i8 sse algorithm to extend half of each 128-bit lane using punpcklbw/punpckhbw. Multiplies all the low half lanes and high half lanes together in separate operations. Then merges the half lane results back together using packuswb.
Unfortunately, some of the cases in vector-reduce-mul.ll regress because we aren't narrowing the vector width of the multiplies as we reduce. The splitting was somewhat making up for that before by causing halves to be discarded after the split.
Differential Revision: https://reviews.llvm.org/D54668
llvm-svn: 347240
This allows to avoid scratch use or indirect VGPR addressing for
small vectors.
Differential Revision: https://reviews.llvm.org/D54606
llvm-svn: 347231
Summary:
This makes it easier/cleaner to generate a single signature from
this directive. Also:
- Adds the symbol name, such that we don't depend on the location
of this directive anymore.
- Actually constructs the signature in the assembler, and make the
assembler own it.
- Refactor the use of MVT vs ValType in the streamer and assembler
to require less conversions overall.
- Changed 700 or so tests to use it.
Reviewers: sbc100, dschuff
Subscribers: jgravelle-google, eraman, aheejin, sunfish, jfb, llvm-commits
Differential Revision: https://reviews.llvm.org/D54652
llvm-svn: 347228
This patch defines an interleaved-load-combine pass. The pass searches
for ShuffleVector instructions that represent interleaved loads. Matches are
converted such that they will be captured by the InterleavedAccessPass.
The pass extends LLVMs capabilities to use target specific instruction
selection of interleaved load patterns (e.g.: ld4 on Aarch64
architectures).
Differential Revision: https://reviews.llvm.org/D52653
llvm-svn: 347208
Truncs are treated as sources if their produce a value of the same
type as the one we currently trying to promote. Truncs used to be
considered as a sink if their operand was the same value type.
We now allow smaller types in the search, so we should search through
truncs that produce a smaller value. These truncs can then be
converted to an AND mask.
This leaves sinks as being:
- points where the value in the register is being observed, such as
an icmp, switch or store.
- points where value types have to match, such as calls and returns.
- zext are included to ease the transformation and are generally
removed later on.
During this change, it also became apart from truncating sinks was
broken: if a sink used a source, its type information had already
been lost by the time the truncation happens. So I've changed the
method of caching the type information.
Differential Revision: https://reviews.llvm.org/D54515
llvm-svn: 347191
There is no variable-length shifts on MSP430. Therefore
"eat" 8 bits of shift via bswap & ext.
Path by Kristina Bessonova!
Differential Revision: https://reviews.llvm.org/D54623
llvm-svn: 347187
The shift requires a copy to avoid clobbering a register. Comparing with 0 uses an xor to produce 0 that will be overwritten with the compare results. So still requires 2 instructions, but should be one byte shorter since it doesn't need to encode an immediate.
llvm-svn: 347185
Previously we used an arithmetic shift right by 31, but that requires a copy to preserve the input. So we might as well materialize a zero and compare to it since the comparison will overwrite the register that contains the zeros. This should be one byte shorter.
llvm-svn: 347181
Leave just the v4i8->v4i64 and v8i8->v8i64, but only enable them on pre-sse4.1 targets when 64-bit mode is enabled. In those cases we end up creating sext loads that get scalarized to code that looks better than what we get from loading into a vector register and doing a multiple step sign extend using unpacks and shifts.
llvm-svn: 347180
Pre-SSE4.1 sext_invec for v2i64 is complicated because we don't have a v2i64 sra instruction. So instead we sign extend to i32 using unpack and sra, then copy the elements and do a v4i32 sra to fill with sign bits, then interleave the i32 sign extend and the sign bits. So really we're doing to two sign extends but only using half of the v4i32 intermediate result.
When the result is more than 128 bits, default type legalization would prefer to split the destination type all the way down to v2i64 with shuffles followed by v16i8/v8i16->v2i64 sext_inreg operations. This results in more instructions than necessary because we are only utilizing the lower 2 elements of the v4i32 intermediate result. Instead we can custom split a v4i8/v4i16->v4i64 sign_extend. Then we can sign extend v4i8/v4i16->v4i32 invec producing a full v4i32 result. Create the sign bit vector as a v4i32 then split and interleave with the sign bits using an punpackldq and punpackhdq.
llvm-svn: 347176
Some of these sequeces look pretty bad since we have to copy the sign bit from a 32 bit register to a 64 bit register to finish a sign extend.
llvm-svn: 347175
If we widen illegal types instead of promoting, we should be able to rely on the type legalizer to create the vector_inreg operations for us with some caveats.
This patch disables combineToExtendVectorInReg when we are using widening.
I've enabled custom legalization for v8i8->v8i64 extends under avx512f since the type legalizer would want to create a vector_inreg with a v64i8 input type which isn't legal without avx512bw. So we go to v16i8 with custom code using the relaxation of rules we get from D54346.
I've also enable custom legalization of v8i64 and v16i32 operations with with AVX. When the input type is 128 bits, the default splitting legalization would extend first 128->256, then do the a split to two 128 pieces. Extend each half to 256 and then concat the result. The custom legalization I've added instead uses a 128->256 bit vector_inreg extend that only reads the lower 64-bits for the low half of the split. Then shuffles the high 64-bits to the low 64-bits and does another vector_inreg extend.
llvm-svn: 347172
Summary: This is an improvement over the two pshufbs and punpcklqdq we'd get otherwise.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54671
llvm-svn: 347171
Sadly, this duplicates (twice) the logic from InstSimplify. There
might be some way to at least share the DAG versions of the code,
but copying the folds seems to be the standard method to ensure
that we don't miss these folds.
Unlike in IR, we don't run DAGCombiner to fixpoint, so there's no
way to ensure that we do these kinds of simplifications unless the
code is repeated at node creation time and during combines.
There were other tests that would become worthless with this
improvement that I changed as pre-commits:
rL347161
rL347164
rL347165
rL347166
rL347167
I'm not sure how to salvage the remaining tests (diffs in this patch).
So the x86 tests verify that the new code is working as intended.
The AMDGPU test is actually similar to my motivating case: we have
some undef value that has survived to machine IR in an x86 test, and
then it gets folded in some weird way, or we crash if we don't transfer
the undef flag. But we would have been better off never getting to that
point by doing these simplifications.
This will lead back to PR32023 someday...
https://bugs.llvm.org/show_bug.cgi?id=32023
llvm-svn: 347170
We were using the 'normalized' shuffle mask from resolveTargetShuffleInputs, which replaces zero/undef inputs with sentinel values. For SimplifyDemandedVectorElts we need the raw mask so we can correctly demand those 'zero' inputs that got normalized away, this requires an extra bit of logic to locally normalize undef inputs.
llvm-svn: 347158
The zero extend will require two stages of unpacks to implement. So its better to shrink the multiply using pmullw and then extend that result back to v4i32 using a single unpack.
llvm-svn: 347149
Comments in llc outputs are printed differently on different
platforms, some with '#', some with '##'. Removed non-essential
part of the checks.
llvm-svn: 347112
This tries to force the result type to vXi32 followed by a truncate. This can help avoid scalarization that would otherwise occur.
There's some annoying examples of an avx512 truncate instruction followed by a packus where we should really be able to just use one truncate. But overall this is still a net improvement.
llvm-svn: 347105
In one case probably you have be using it, in the other it
looks like it was redundant.
Differential Revision: https://reviews.llvm.org/D54644
llvm-svn: 347098
This NFC patch just adds test cases for conversions that currently
require scalarization of vectors. An updcoming patch will change
the legalization for these and it is more suitable on the review
to show the diferences in code gen rather than just the new code gen.
llvm-svn: 347090
When unwinding past a function that uses shadow call stack, we must
subtract 8 from the value of the x18 register. This patch causes us
to emit a call frame instruction that causes that to happen.
Differential Revision: https://reviews.llvm.org/D54609
llvm-svn: 347089
Summary:
As discussed in previous review, and noted in the FIXME, if `X` is actually an `lshr Y, Z` (logical!),
we can fold the `Z` into 'control`, and let the `BEXTR` do this too.
We could just insert those 8 bits of shift amount into control,
but it is better to instead zero-extend them, and 'or' them in place.
We can only do this for `lshr`, not `ashr`, because we do not know that the mask cover only the bits of `Y`,
and not any of the sign-extended bits.
The obvious question is, is this actually legal to do?
I believe it is. Relevant quotes, from `Intel® 64 and IA-32 Architectures Software Developer’s Manual`, `BEXTR — Bit Field Extract`:
* `Bit 7:0 of the second source operand specifies the starting bit position of bit extraction.`
* `A START value exceeding the operand size will not extract any bits from the second source operand.`
* `Only bit positions up to (OperandSize -1) of the first source operand are extracted.`
* `All higher order bits in the destination operand (starting at bit position LENGTH) are zeroed.`
* `The destination register is cleared if no bits are extracted.`
FIXME: if we can do this, i wonder if we should prefer `BEXTR` over `BZHI` in such cases.
Reviewers: RKSimon, craig.topper, spatel, andreadb
Reviewed By: RKSimon, craig.topper, andreadb
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54095
llvm-svn: 347048
This commit introduces support for materialising 64-bit constants for RV64I,
making use of the RISCVMatInt::generateInstSeq helper in order to share logic
for immediate materialisation with the MC layer (where it's used for the li
pseudoinstruction).
test/CodeGen/RISCV/imm.ll is updated to test RV64, and gains new 64-bit
constant tests. It would be preferable if anyext constant returns were sign
rather than zero extended (see PR39092). This patch simply adds an explicit
signext to the returns in imm.ll.
Further optimisations for constant materialisation are possible, most notably
for mask-like values which can be generated my loading -1 and shifting right.
A future patch will standardise on the C++ codepath for immediate selection on
RV32 as well as RV64, and then add further such optimisations to
RISCVMatInt::generateInstSeq in order to benefit both RV32 and RV64 for
codegen and li expansion.
Differential Revision: https://reviews.llvm.org/D52962
llvm-svn: 347042
By early promoting the multiply to use an i16 element type we can avoid op legalization emit a second multiply for the 8 upper elements of the v16i8 type we would otherwise get.
llvm-svn: 347032
If a block had one of the _term instructions used for gluing
exec modifying instructions to the end of the block,
analyzeBranch would fail, preventing the verifier from catching
a broken successor list.
llvm-svn: 347027
We aren't going to use the upper bits of the multiply result that the extend would effect. So we don't need a specific type of extend.
This makes some reduction test cases shorter because we were previously trying to sign_extend a truncate which we can't eliminate.
llvm-svn: 347011
Add a pass to fixup various vector ISel issues.
Currently we handle converting GLOBAL_{LOAD|STORE}_*
and GLOBAL_Atomic_* instructions into their _SADDR variants.
This involves feeding the sreg into the saddr field of the new instruction.
llvm-svn: 347008
Summary:
`throw` instruction is a terminator in wasm, but BBs were not splitted
after `throw` instructions, causing machine instruction verifier to
fail.
This patch
- Splits BBs after `throw` instructions in WasmEHPrepare and adding an
unreachable instruction after `throw`, which will be deleted in
LateEHPrepare pass
- Refactors WasmEHPrepare into two member functions
- Changes the semantics of `eraseBBsAndChildren` in LateEHPrepare pass
to match that of WasmEHPrepare pass, which is newly added. Now
`eraseBBsAndChildren` does not delete BBs with remaining predecessors.
- Fixes style nits, making static function names conform to clang-tidy
- Re-enables the test temporarily disabled by rL346840 && rL346845
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54571
llvm-svn: 347003
In reduceVMULWidth, we no longer need to worry about extending the vector to 128 bits first. Regular widening of extends, muls and shuffles will take care of that for us.
In combineMulToPMADDWD, we can handle v2i32 multiplies and allow the VPMADDWD to be widened to v4i32 during type legalization by adding custom widening like we do have for AVG/ADDUS/SUBUS. I had to modify that code a little to allow different and output VTs.
Differential Revision: https://reviews.llvm.org/D54512
llvm-svn: 346980
This fixes -filetype=null support when compiling for a Win32 target and the module has a CodeView flag.
The only places changed are the uses of getTargetStreamer function - this patch guards both of them with null checks.
Committed on behalf of @eush (Eugene Sharygin)
Differential Revision: https://reviews.llvm.org/D54008
llvm-svn: 346962
Mark the FREM SelectionDAG node as Expand, which is necessary in order to
support the frem IR instruction on RISC-V. This is expanded into a library
call. Adds the corresponding test. Previously, this would have triggered an
assertion at instruction selection time.
Differential Revision: https://reviews.llvm.org/D54159
Patch by Luís Marques.
llvm-svn: 346958
Reapply r346374 with the fixes for modules build.
Original summary:
This change implements assembler parser, code emitter, ELF object writer
and disassembler for the MSP430 ISA. Also, more instruction forms are added
to the target description.
Patch by Michael Skvortsov!
llvm-svn: 346948
This avoids some nasty shuffles when we have avx512. It will also prevent using zmm truncate instructions when a ymm instruction that zeroes part of an xmm register will do. Also avoid using avx512 truncate instructions when the input is 128 bits or less. These instructions are 2 uops on skx so we can probably find a better single uop shuffle like pshufb.
llvm-svn: 346936
The narrow types end up requesting widening, but generic legalization will end up scalaring and using a build_vector to do the widening.
llvm-svn: 346916
On 64-bit targets the type legalizer will use i64 to legalize these. But when i64 isn't legal, the type legalizer won't try an FP type. So do it manually instead.
There are a few regressions in here due to some v2i32 operations like mul and div now being reassembled into a full vector just to store instead of storing the pieces. But this was already occuring in 64-bit mode so its not a new issue.
llvm-svn: 346908
The machine scheduler currently biases register copies to/from
physical registers to be closer to their point of use / def to
minimize their live ranges. This change extends this to also physical
register assignments from immediate values.
This causes a reduction in reduction in overall register pressure and
minor reduction in spills and indirectly fixes an out-of-registers
assertion (PR39391).
Most test changes are from minor instruction reorderings and register
name selection changes and direct consequences of that.
Reviewers: MatzeB, qcolombet, myatsina, pcc
Subscribers: nemanjai, jvesely, nhaehnle, eraman, hiraditya,
javed.absar, arphaman, jfb, jsji, llvm-commits
Differential Revision: https://reviews.llvm.org/D54218
llvm-svn: 346894
Narrower vectors will be widened to 128 bits without changing the element size. And generic type legalization can already handle widening mulhu/mulhs.
Differential Revision: https://reviews.llvm.org/D54513
llvm-svn: 346879
This patch removes the last use of the constant pool shuffle decode helper and consistently uses the 'getTargetShuffleMaskIndices' versions instead. The constant pool versions are now purely used for assembly comments.
The avx512vbmi intrinsic upgrades had to be altered as they were being decoded as broadcasts, similar to what I fixed in rL346032. I don't think the change is critical - although its annoying that we lose the {k}{z} instruction test coverage as they are tricky to generate....
Differential Revision: https://reviews.llvm.org/D54083
llvm-svn: 346850
I've only added sse2 and sse4.1 variants as I'm only interested in the two v4i16 tests and I don't expect that to different with AVX other than a v prefix.
llvm-svn: 346834
Summary:
This adds support for the 'event section' specified in the exception
handling proposal. (This was named 'exception section' first, but later
renamed to 'event section' to take possibilities of other kinds of
events into consideration. But currently we only store exception info in
this section.)
The event section is added between the global section and the export
section. This is for ease of validation per request of the V8 team.
This patch:
- Creates the event symbol type, which is a weak symbol
- Makes 'throw' instruction take the event symbol '__cpp_exception'
- Adds relocation support for events
- Adds WasmObjectWriter / WasmObjectFile (Reader) support
- Adds obj2yaml / yaml2obj support
- Adds '.eventtype' printing support
Reviewers: dschuff, sbc100, aardappel
Subscribers: jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54096
llvm-svn: 346825
To make ISD::VSELECT available(legal) so long as there are altivec instruction, otherwise it's default behavior is expanding,
which is legalized at type-legalization phase. Use xxsel to match vselect if vsx is open, or use vsel.
Differential Revision: https://reviews.llvm.org/D49531
llvm-svn: 346824
The scan was incorrectly skipping the first instruction, so a register
could appear to be dead when it was actually live. This eventually leads
to a machine verifier failure and miscompile in arm-ldst-opt.
Differential Revision: https://reviews.llvm.org/D54491
llvm-svn: 346821
An extractelement with non-constant index will be lowered either to
scratch or movrel loop in most cases. This patch converts such
instruction into a set of selects if vector size is not too big.
Differential Revision: https://reviews.llvm.org/D54351
llvm-svn: 346800
Legalizer used to request an ext load from i8 to i1 when promoting
vector element type to i8. Fixed.
Differential Revision: https://reviews.llvm.org/D54440
llvm-svn: 346795
The IEEE-754 Standard makes it clear that fneg(x) and
fsub(-0.0, x) are two different operations. The former is a bitwise
operation, while the latter is an arithmetic operation. This patch
creates a dedicated FNeg IR Instruction to model that behavior.
Differential Revision: https://reviews.llvm.org/D53877
llvm-svn: 346774
Usually MIPS hosts uses a legacy (non IEEE 754-2008) encoding for NaNs.
Tests like `nan_f32` failed in attempt to compare hard-coded IEEE
754-2008 NaN value and a legacy NaN value provided by a system.
llvm-svn: 346773
If a loaded value is replicated it is best to combine these two operations
into a VLREP (load and replicate), but isel will not produce this if the load
has other users as well.
This patch handles this by putting the other users of the load to use the
REPLICATE 0-element instead of the load. This way the load has only the
REPLICATE node as user, and we get a VLREP.
Review: Ulrich Weigand
https://reviews.llvm.org/D54264
llvm-svn: 346746
I'm looking into whether we can make this the default legalization strategy. Adding these tests to help cover the changes that will be necessary.
This patch adds copies of some tests with the command line switch enabled. By making copies its easier to compare the two legalization strategies.
I've also removed RUN lines from some of these tests that already had -x86-experimental-vector-widening-legalization
llvm-svn: 346745
This patch adds the ability to use a PALIGNR to rotate a pair of inputs to select a range containing all the referenced elements, followed by a single input permute to put them in the right location.
Differential Revision: https://reviews.llvm.org/D54267
llvm-svn: 346706
Summary:
This is to replace the ELFAsmParser that WebAssembly was using, which
so far was a stub that didn't do anything, and couldn't work correctly
with wasm.
This new class is there to implement generic directives related to
wasm as a binary format. Wasm target specific directives are still
parsed in WebAssemblyAsmParser as before. The two classes now
cooperate more correctly too.
Also implemented .result which was missing. Any unknown directives
will now result in errors.
Reviewers: dschuff, sbc100
Subscribers: mgorny, jgravelle-google, eraman, aheejin, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54360
llvm-svn: 346700
Truncate and shuffle lowering are already capable of matching to PACKUS using known bits analysis.
This features one test change where we now prefer to extend v16i16->v16i32 then trunc v16i32->v16i8 over extract_subvector+packus when avx512f is available, but avx512bw is not.
llvm-svn: 346697
Sometimes after basic block placement we end up with a code like:
sreg = s_mov_b64 -1
vcc = s_and_b64 exec, sreg
s_cbranch_vccz
This happens as a join of a block assigning -1 to a saved mask and
another block which consumes that saved mask with s_and_b64 and a
branch.
This is essentially a single s_cbranch_execz instruction when moved
into a single new basic block.
Differential Revision: https://reviews.llvm.org/D54164
llvm-svn: 346690
The function only checks that instruction reads a super-register
containing requested physical register. In case if a sub-register
if being read that is also a use of a super-reg, so added the check.
In particular MI->readsRegister() is broken because of the missing
check. The resulting check is essentially regsOverlap().
Differential Revision: https://reviews.llvm.org/D54128
llvm-svn: 346686
This extends the .option support from D45864 to enable/disable the relax
feature flag from D44886
During parsing of the relax/norelax directives, the RISCV::FeatureRelax
feature bits of the SubtargetInfo stored in the AsmParser are updated
appropriately to reflect whether relaxation is currently enabled in the
parser. When an instruction is parsed, the parser checks if relaxation is
currently enabled and if so, gets a handle to the AsmBackend and sets the
ForceRelocs flag. The AsmBackend uses a combination of the original
RISCV::FeatureRelax feature bits set by e.g -mattr=+/-relax and the
ForceRelocs flag to determine whether to emit relocations for symbol and
branch diffs. Diff relocations should therefore only not be emitted if the
relax flag was not set on the command line and no instruction was ever parsed
in a section with relaxation enabled to ensure correct diffs are emitted.
Differential Revision: https://reviews.llvm.org/D46423
Patch by Lewis Revill.
llvm-svn: 346655
Summary:
Handle extra output from index loads in cases where we wish to
forward a load value directly from a preceeding store.
Fixes PR39571.
Reviewers: peter.smith, rengolin
Subscribers: javed.absar, hiraditya, arphaman, llvm-commits
Differential Revision: https://reviews.llvm.org/D54265
llvm-svn: 346654
Iterate over all elements and count the number of uses among them for each
used load. Then make sure to REPLICATE the load which has the most uses in
order to minimize the number of needed element insertions.
Review: Ulrich Weigand
https://reviews.llvm.org/D54322
llvm-svn: 346637
This is a long-awaited follow-up suggested in D33578. Since then, we've picked up even more
opportunities for vector narrowing from changes like D53784, so there are a lot of test diffs.
Apart from 2-3 strange cases, these are all wins.
I've structured this to be no-functional-change-intended for any target except for x86
because I couldn't tell if AArch64, ARM, and AMDGPU would improve or not. All of those
targets have existing regression tests (4, 4, 10 files respectively) that would be
affected. Also, Hexagon overrides the shouldReduceLoadWidth() hook, but doesn't show
any regression test diffs. The trade-off is deciding if an extra vector load is better
than a single wide load + extract_subvector.
For x86, this is almost always better (on paper at least) because we often can fold
loads into subsequent ops and not increase the official instruction count. There's also
some unknown -- but potentially large -- benefit from using narrower vector ops if wide
ops are implemented with multiple uops and/or frequency throttling is avoided.
Differential Revision: https://reviews.llvm.org/D54073
llvm-svn: 346595
There are two AGU units, and per 1cy, there can be either two loads,
or a load and a store; but not two stores, or two loads and a store.
Additionally, loads shouldn't affect the store scheduler and vice versa.
(but *should* affect the PdEX scheduler.)
Required rL346545.
Fixes https://bugs.llvm.org/show_bug.cgi?id=39465
llvm-svn: 346587
The sdivrem will emit its own MOVSX to move %ah to the low byte of a register. By using a MOVSX for an any_extend this allows a post-isel peephole to merge them.
llvm-svn: 346581
After the division %ah is being sign extended to move it to lower byte of a register while avoiding a partial register read. We then zero extend the low byte to the full 32 bit register. But we don't use any of the zero extended bits. In the DAG the zero extend was really an any_extend so the sign extend should have been enough.
llvm-svn: 346580
This gives shuffle lowering the freedom to use zero_extend_vector_inreg for the unpckl shuffle. Shuffle combining usually makes this swap later, but not when AVX512 is enabled it seems.
While there also use DAG.getConstant to create a 0 vector instead of using the helper the forces a specific BUILD_VECTOR. I don't think that helper is usually needed. We're basically free to create a constant build_vector anytime and it will be legalized on its own.
llvm-svn: 346574
This patch adds support for funclets in frame lowering and ISel
lowering. Together with D50288 and D50166, it enables C++ exception
handling.
Patch by Sanjin Sijaric, with some fixes by me.
Differential Revision: https://reviews.llvm.org/D51524
llvm-svn: 346568
In r346432 ("[DAGCombine] Improve alias analysis for chain of independent stores"),
the order of ldi/sts blocks changed.
The new IR is equivalent to the old IR.
This patch updates the test to fix the test suite.
llvm-svn: 346565
LDRcp should be deleted when the dest register is dead in register
coalescing. Without MemOp, dead LDRcp will cause dead constant pool
value which references to non-existing label.
Patch by Yin Ma.
Differential Revision: https://reviews.llvm.org/D54173
llvm-svn: 346563
Summary:
These tests fail on 32-bit builds because NaN payload bits in floating point
immediates are not necessarily preserved through compilation. This is because
the MC layer uses native doubles to store these values. The tests will be
reenabled once this problem has been fixed or deleted if we decide we don't care
about lowering payload bits.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54353
llvm-svn: 346558
With avx512f but not avx512bw we need to extend to v16i32 then truncate that to to v16i8. Previously we emitted both nodes during lowering, but I'm trying to switch to using target independent nodes and with that switched the extend+truncate wou
This patch changes the implementation to what will be necessary with that patch which helps minimize test diffs.
llvm-svn: 346552
A few fp128 tests were omitted from test/CodeGen/SystemZ/fp-round-01.ll
since in early days, LLVM couldn't handle implicitly generated library
calls to functions with long double arguments on SystemZ.
This deficiency was actually long since fixed, but those tests are
still missing. This patch adds the missing tests. NFC.
llvm-svn: 346541
This makes X86ISD::VSEXT more similar to ISD::SIGN_EXTEND and ISD::ZERO_EXTEND.
I'm hoping to replace X86ISD::VSEXT/VZEXT with target independent nodes. Making the target specific nodes similar to the target independent nodes helps minimize test diffs in that patch.
llvm-svn: 346539
Eliminate the stack frame in functions with the noreturn nounwind
attributes, and when the noreturn-stack-elim target feature is
enabled. This reduces the code and stack space needed for noreturn
functions.
Differential Revision: https://reviews.llvm.org/D54210
llvm-svn: 346532
It's possible for vector op legalization to generate a shuffle. If that happens we should give a chance for DAG combine to combine that with a build_vector input.
I also fixed a bug in combineShuffleOfScalars that was considering the number of uses on a undef input to a shuffle. We don't care how many times undef is used.
Differential Revision: https://reviews.llvm.org/D54283
llvm-svn: 346530
Both -fPIC and -G0 disable placement of globals in small data section,
but if a global has an explicit section assigmnent placing it in small
data, it should go there anyway.
llvm-svn: 346523
Currently in llvm, CalleeSavedInfo can only assign a callee saved register to
stack frame index to be spilled in the prologue. We would like to enable
spilling gprs to vector registers. This patch adds the capability to spill to
other registers aside from just the stack. It also adds the changes for power9
to spill gprs to volatile vector registers when they are available.
This happens only for leaf functions when using the option
-ppc-enable-pe-vector-spills.
Differential Revision: https://reviews.llvm.org/D39386
llvm-svn: 346512
A minor improvement of buildVector() that skips creating an
INSERT_VECTOR_ELT for a Value which has already been used for the
REPLICATE.
Review: Ulrich Weigand
https://reviews.llvm.org/D54315
llvm-svn: 346504
Now that we have mixed type sizes, i1 values need to be explicitly
handled as we want to avoid promoting these values.
Differential Revision: https://reviews.llvm.org/D54308
llvm-svn: 346499
I noticed that we weren't generating broadcasts as much I thought we would with
D54271, and this is part of the problem.
Widening the shuffle elements means adding bitcasts and hiding the relationship
between a splatted scalar and the vector. If we can form a broadcast, do that
before going through the rest of the shuffle lowering because broadcasts should
be cheap and can often be load-folded.
Differential Revision: https://reviews.llvm.org/D54280
llvm-svn: 346498
The DAGCombiner tries to SimplifySelectCC as follows:
select_cc(x, y, 16, 0, cc) -> shl(zext(set_cc(x, y, cc)), 4)
It can't cope with the situation of reordered operands:
select_cc(x, y, 0, 16, cc)
In that case we just need to swap the operands and invert the Condition Code:
select_cc(x, y, 16, 0, ~cc)
Differential Revision: https://reviews.llvm.org/D53236
llvm-svn: 346484
In SimplifyCFG when given a conditional branch that goes to BB1 and BB2, the hoisted common terminator instruction in the two blocks, caused debug line records associated with subsequent select instructions to become ambiguous. It causes the debugger to display unreachable source lines.
Differential Revision: https://reviews.llvm.org/D53390
llvm-svn: 346481
Previously, during the search, all values had to have the same
'TypeSize', which is equal to number of bits of the integer type of
the icmp operand. All values in the tree had to match this size;
meaning that, if we searched from i16, we wouldn't accept i8s. A
change in type size requires zext and truncs to perform the casts so,
to allow mixed narrow types, the handling of these instructions is
now slightly different:
- we allow casts if their result or operand is <= TypeSize.
- zexts are sinks if their result > TypeSize.
- truncs are still sinks if their operand == TypeSize.
- truncs are still sources if their result == TypeSize.
The transformation bails on finding an icmp that operates on data
smaller than the current TypeSize.
Differential Revision: https://reviews.llvm.org/D54108
llvm-svn: 346480
Summary:
The pass incorrectly assumed if there's a longjmp declaration in the
module, there is also a setjmp function declaration. Fixed it, and now
the pass only converts longjmp and does not do any other transformation
when there's no setjmp declaration in the module.
Fixes PR39562.
Reviewers: jgravelle-google, sbc100
Subscribers: dschuff, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54273
llvm-svn: 346445
As discussed in D54073, we have a potential regression from more aggressive vector narrowing here, so let's try to avoid that by changing build-vector lowering slightly.
Insert-vector-element lowering always does this since there's no "pinsr" for ymm/zmm:
// If the vector is wider than 128 bits, extract the 128-bit subvector, insert
// into that, and then insert the subvector back into the result.
...but we can sometimes do better for insert-into-constant-vector by using shuffle lowering.
Differential Revision: https://reviews.llvm.org/D54271
llvm-svn: 346433
FindBetterNeighborChains simulateanously improves the chain
dependencies of a chain of related stores avoiding the generation of
extra token factors. For chains longer than the GatherAllAliasDepths,
stores further down in the chain will necessarily fail, a potentially
significant waste and preventing otherwise trivial parallelization.
This patch directly parallelize the chains of stores before improving
each store. This generally improves DAG-level parallelism.
Reviewers: courbet, spatel, RKSimon, bogner, efriedma, craig.topper, rnk
Subscribers: sdardis, javed.absar, hiraditya, jrtc27, atanasyan, llvm-commits
Differential Revision: https://reviews.llvm.org/D53552
llvm-svn: 346432
It was discovered in randomized testing that the SystemZ implementation of
shouldCoalesce() could be caused to crash when subreg liveness was
enabled. This was because an undef use of the virtual register was copied
outside current MBB at the point of shouldCoalesce() being called. For more
details, see https://bugs.llvm.org/show_bug.cgi?id=39276.
This patch changes the check for MBB locality from livein/liveout checks to
do checks for all instructions of both intervals being inside MBB. This
avoids the cases with dead defs / undef uses outside MBB, which are not
affecting liveness in/out of MBB.
The original test case included as a reduced .mir test case.
Review: Ulrich Weigand
https://reviews.llvm.org/D54197
llvm-svn: 346406
Generalize code in Thumb2InstrInfo::storeRegToStackSlot() and
loadRegToStackSlot() to allow the GPR class or any of its sub-classes
(including hGPR) to be stored/loaded by ARM::t2STRi12/ARM::t2LDRi12.
Differential Revision: https://reviews.llvm.org/D51927
llvm-svn: 346401
Promote alloca can vectorize a small array by bitcasting it to a
vector type. Extend vectorization for the case when alloca is
already a vector type. We still want to replace GEPs with an
insert/extract element instructions in this case.
Differential Revision: https://reviews.llvm.org/D54219
llvm-svn: 346376
Summary:
This change implements assembler parser, code emitter, ELF object writer
and disassembler for the MSP430 ISA. Also, more instruction forms are added
to the target description.
Reviewers: asl
Reviewed By: asl
Subscribers: pftbest, krisb, mgorny, llvm-commits
Differential Revision: https://reviews.llvm.org/D53661
llvm-svn: 346374
Like the comment says, this isn't the most efficient fix in terms of
codesize, but it works.
Differential Revision: https://reviews.llvm.org/D54129
llvm-svn: 346358
The lowering was missing live-ins in certain cases, like a sequence of
multiple tMOVCCr_pseudo instructions. This would lead to a verifier
failure, and on pre-v6 Thumb CPSR would be incorrectly clobbered.
For reasons I don't completely understand, it's hard to get a sequence
of multiple tMOVCCr_pseudo instructions; the issue only seems to show up
with 64-bit comparisons where the result is zero-extended. I added some
extra testcases in case that changes in the future. Probably some
optimization opportunities here if anyone is interested. (@test_slt_not
is the case that was getting miscompiled.)
The code to check the liveness of CPSR was stolen from
X86ISelLowering.cpp; maybe it could be refactored into common helper,
but I have no idea where to put it.
Differential Revision: https://reviews.llvm.org/D54192
llvm-svn: 346355
This allows testing AMDGPU alias analysis like any
other alias analysis pass. This fixes the existing
test pointlessly running opt -O3 when it really
just wants to run the one analysis.
Before there was no way to test this using -aa-eval
with opt, since the default constructed pass
is run. The wrapper subclass allows the
default constructor to pass the necessary callback.
llvm-svn: 346353
Summary:
The conditional branch created to support -fsplit-stack for X86 is
left unbiased/unhinted, resulting in less than ideal block placement:
the __morestack call block is kept on the main hot path. Bias the
branch to insure that the stack allocation block is treated as a
"cold" block during machine basic block placement.
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54123
llvm-svn: 346336
It seems that the PPC backend croaks when lowering a call to a
function with an argument of type [2 x i32].
Just modify the type slightly to avoid this -- I wasn't actually
intending to stress test the backend...
llvm/lib/Target/PowerPC/PPCISelLowering.cpp:6172: llvm::SDValue llvm::PPCTargetLowering::LowerCall_64SVR4(...): Assertion `(!HasParameterArea || NumBytesActuallyUsed == ArgOffset) && "mismatch in size of parameter area"' failed.
llvm-svn: 346334
This adds the llvm-side support for post-inlining evaluation of the
__builtin_constant_p GCC intrinsic.
Also fixed SCCPSolver::visitCallSite to not blow up when seeing a call
to a function where canConstantFoldTo returns true, and one of the
arguments is a struct.
Updated from patch initially by Janusz Sobczak.
Differential Revision: https://reviews.llvm.org/D4276
llvm-svn: 346322
Set operands order for G_MERGE_VALUES and G_UNMERGE_VALUES so
that least significant bits always go first, regardless of endianness.
Differential Revision: https://reviews.llvm.org/D54098
llvm-svn: 346305
Set `LiveReg::PhysReg` to zero when freeing a register instead of
removing it from the entry from `LiveRegMap`. This way no iterators get
invalidated and we can avoid passing around and updating iterators all
over the place.
This does not change any allocator decisions. It is not completely NFC
because the arbitrary iteration order through `LiveRegMap` in
`spillAll()` changes so we may get a different order in those spill
sequences (the amount of spills does not change).
This is in preparation of https://reviews.llvm.org/D52010.
llvm-svn: 346298
Add this option for debugging and providing workaround.
By default it is off so no behavior change in backend.
Differential Revision: https://reviews.llvm.org/D54158
llvm-svn: 346267
It was causing a crash because we were trying to get the definition
of a target register. Fixed the issue by adding a check and added
a test case for that.
llvm-svn: 346251
This patch makes LICM use `ICFLoopSafetyInfo` that is a smarter version
of LoopSafetyInfo that leverages power of Implicit Control Flow Tracking
to keep track of throwing instructions and give less pessimistic answers
to queries related to throws.
The ICFLoopSafetyInfo itself has been introduced in rL344601. This patch
enables it in LICM only.
Differential Revision: https://reviews.llvm.org/D50377
Reviewed By: apilipenko
llvm-svn: 346201
This reverts rL345880. It caused some test failures on the
webassembly waterfall. e.g. binaryen2.test_mainenv fails due
the fact that `envp` ends up being undef rather than 0.
Differential Revision: https://reviews.llvm.org/D54117
llvm-svn: 346187
I'm preparing a patch to avoid creating critical edges in cmov expansion. Updating these tests to make the changes by the next patch easier to see.
llvm-svn: 346161
On Power9, we don't have patterns to select the following intrinsics:
llvm.ppc.vsx.stxvw4x.be
llvm.ppc.vsx.stxvd2x.be
This patch adds support for these.
Differential Revision: https://reviews.llvm.org/D53581
llvm-svn: 346148
The constrained intrinsic tests have grown in number. Split off
the FMA tests into their own file to reduce double coverage.
Differential Revision: https://reviews.llvm.org/D53932
llvm-svn: 346137
Expand on LONG_BRANCH_LUi and LONG_BRANCH_(D)ADDiu pseudo
instructions by creating variants which support
less operands/accept GPR64Opnds as their operand in order
to appease the machine verifier pass.
Differential Revision: https://reviews.llvm.org/D53977
llvm-svn: 346133
The new atomic optimizer I previously added in D51969 did not work
correctly when a pixel shader was using derivatives, and had helper
lanes active.
To fix this we add an llvm.amdgcn.ps.live call that guards a branch
around the entire atomic operation - ensuring that all helper lanes are
inactive within the wavefront when we compute our atomic results.
I've added a test case that can cause derivatives, and exposes the
problem.
Differential Revision: https://reviews.llvm.org/D53930
llvm-svn: 346128
Turn the assert in PrepareConstants into a conditon so that we can
handle mul instructions with negative immediates.
Differential Revision: https://reviews.llvm.org/D54094
llvm-svn: 346126
r345840 slightly changed the way promotion happens which could
result in zext and truncs having the same source and destination
types. This fixes that issue.
We can now also remove the zext and trunc in the following case:
(zext (trunc (promoted op)), i32)
This means that we can no longer treat a value, that is only used by
a sink, to be safe to promote.
I've also added in some extra asserts and replaced a cast for a
dyn_cast.
Differential Revision: https://reviews.llvm.org/D54032
llvm-svn: 346125
This patch fixes a bug in the AVR FRMIDX expansion logic.
The expansion would leave a leftover operand from the original FRMIDX,
but now attached to a MOVWRdRr instruction. The MOVWRdRr instruction
did not expect this operand and so LLVM rejected the machine
instruction.
This would trigger an assertion:
Assertion failed: ((isImpReg || Op.isRegMask() || MCID->isVariadic() ||
OpNo < MCID->getNumOperands() || isMetaDataOp) &&
"Trying to add an operand to a machine instr that is already done!"),
function addOperand, file llvm/lib/CodeGen/MachineInstr.cpp
Tim fixed this so that now the FRMIDX is expanded correctly into
a well-formed MOVWRdRr.
Patch by Tim Neumann
llvm-svn: 346117
v2i8/v2i16/v2i32 are promoted to v2i64. pmuludq takes a v2i64 input and produces a v2i64 output. Since we don't about the upper bits of the type legalized multiply we can use the pmuludq to produce the multiply result for the bits we do care about.
llvm-svn: 346115
This is an AVR-specific workaround for a limitation of the register
allocator that only exposes itself on targets with high register
contention like AVR, which only has three pointer registers.
The three pointer registers are X, Y, and Z.
In most nontrivial functions, Y is reserved for the frame pointer,
as per the calling convention. This leaves X and Z. Some instructions,
such as LPM ("load program memory"), are only defined for the Z
register. Sometimes this just leaves X.
When the backend generates a LDDWRdPtrQ instruction with Z as the
destination pointer, it usually trips up the register allocator
with this error message:
LLVM ERROR: ran out of registers during register allocation
This patch is a hacky workaround. We ban the LDDWRdPtrQ instruction
from ever using the Z register as an operand. This gives the
register allocator a bit more space to allocate, fixing the
regalloc exhaustion error.
Here is a description from the patch author Peter Nimmervoll
As far as I understand the problem occurs when LDDWRdPtrQ uses
the ptrdispregs register class as target register. This should work, but
the allocator can't deal with this for some reason. So from my testing,
it seams like (and I might be totally wrong on this) the allocator reserves
the Z register for the ICALL instruction and then the register class
ptrdispregs only has 1 register left and we can't use Y for source and
destination. Removing the Z register from DREGS fixes the problem but
removing Y register does not.
More information about the bug can be found on the avr-rust issue
tracker at https://github.com/avr-rust/rust/issues/37.
A bug has raised to track the removal of this workaround and a proper
fix; PR39553 at https://bugs.llvm.org/show_bug.cgi?id=39553.
Patch by Peter Nimmervoll
llvm-svn: 346114
Summary: This also enables some constant folding from KnownBits propagation. This helps on some cases vXi64 case in 32-bit mode where constant vectors appear as vXi32 and a bitcast. This can prevent getNode from constant folding sra/shl/srl.
Reviewers: RKSimon, spatel
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54069
llvm-svn: 346102
Use MachineFrameInfo's OffsetAdjustment field to pass this information
from the target to CodeViewDebug.cpp. The X86 backend doesn't use it for
any other purpose.
This fixes PR38857 in the case where there is a non-aligned quantity of
CSRs and a non-aligned quantity of locals.
llvm-svn: 346062
The majority of the changes are because the rest of shuffle lowering/combining prefers to replace the undef input with the other operand. Using UNPCKL directly seemed to avoid this and just grabbed a randomish register for the undef which can create false dependencies.
llvm-svn: 346050
We already have custom lowering for the AVX case in LegalizeVectorOps. So its better to keep the regular extend op around as long as possible.
I had to qualify one place in DAG combine that created illegal vector extending load operations. This change by itself had no effect on any tests which is why its included here.
I've made a few cleanups to the custom lowering. The sign extend code no longer creates an identity shuffle with undef elements. The zero extend code now emits a zero_extend_vector_inreg instead of an unpckl with a zero vector.
For the high half of the custom lowering of zero_extend/any_extend, we're now using an unpckh with a zero vector or undef. Previously we used used a pshufd to move the upper 64-bits to the lower 64-bits and then used a zero_extend_vector_inreg. I think the zero vector should require less execution resources and be smaller code size.
Differential Revision: https://reviews.llvm.org/D54024
llvm-svn: 346043
A number of intrinsics, such as llvm.sin.f32, would result in a failure to
select. This patch adds expansions for the relevant selection DAG nodes, as
well as exhaustive testing for all f32 and f64 intrinsics.
The codegen for FMA remains a TODO item, pending support for the various
RISC-V FMA instruction variants.
The llvm.minimum.f32.* and llvm.maximum.* tests are commented-out, pending
upstream support for target-independent expansion, as discussed in
http://lists.llvm.org/pipermail/llvm-dev/2018-November/127408.html.
Differential Revision: https://reviews.llvm.org/D54034
Patch by Luís Marques.
llvm-svn: 346034
This is necessary as I'm wanting to remove the 'Constant Pool' shuffle decoding from getTargetShuffleMask - but using getTargetShuffleMaskIndices allows the shuffle combiner to realize that these calls are really broadcasts.....
As with a lot of the X86ISD::VPERMV3 code this causes some vperm2i/vperm2t shuffles to flip depending on optimal commutation.
llvm-svn: 346032
Summary:
EH stack depth is incremented at `try` and decremented at `catch`. When
there are more than two catch instructions for a try instruction, we
shouldn't count non-first catches when calculating EH stack depths.
This patch fixes two bugs:
- CFGStackify: Exclude `catch_all` in the terminate catch pad when
calculating EH pad stack, because when we have multiple catches for a
try we should count only the first catch instruction when calculating
EH pad stack.
- InstPrinter: The initial intention was also to exclude non-first
catches, but it didn't account nested try-catches, so it failed on
this case:
```
try
try
catch
end
catch <-- (1)
end
```
In the example, when we are at the catch (1), the last seen EH
instruction is not `try` but `end_try`, violating the wrong assumption.
We don't need these after we switch to the second proposal because there
is gonna be only one `catch` instruction. But anyway before then these
bugfixes are necessary for keep trunk in working state.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D53819
llvm-svn: 346029
As reported in PR38952, postra-machine-sink relies on DBG_VALUE insns being
adjacent to the def of the register that they reference. This is not always
true, leading to register copies being sunk but not the associated DBG_VALUEs,
which gives the debugger a bad variable location.
This patch collects DBG_VALUEs as we walk through a BB looking for copies to
sink, then passes them down to performSink. Compile-time impact should be
negligable.
Differential Revision: https://reviews.llvm.org/D53992
llvm-svn: 345996
Small-data (i.e. GP-relative) loads and stores allow 16-bit scaled
offset. For a load of a value of type T, the small-data area is
equivalent to an array "T sdata[65536]". This implies that objects
of smaller sizes need to be closer to the beginning of sdata,
while larger objects may be farther away, or otherwise the offset
may be insufficient to reach it. Similarly, an object of a larger
size should not be accessed via a load of a smaller size.
llvm-svn: 345975
reduceBuildVecConvertToConvertBuildVec vectorizes int2float in the DAGCombiner, which means that even if the LV/SLP has decided to keep scalar code using the cost models, this will override this.
While there are cases where vectorization is necessary in the DAG (mainly due to legalization artefacts), I don't think this is the case here, we should assume that the vectorizers know what they are doing.
Differential Revision: https://reviews.llvm.org/D53712
llvm-svn: 345964
Summary:
Assembly output can use globals like __stack_pointer implicitly,
but has no way of indicating the type of such a global, which makes
it hard for tools processing it (such as the MC Assembler) to
reconstruct this information.
The improved assembler directives parsing (in progress in
https://reviews.llvm.org/D53842) will make use of this information.
Also deleted code for the .import_global directive which was unused.
New test case in userstack.ll
Reviewers: dschuff, sbc100
Subscribers: jgravelle-google, aheejin, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54012
llvm-svn: 345917
I'm having trouble creating a test case for the ISD::TRUNCATE part of this that shows any codegen differences. But I was able to test the setcc path which is what the test changes here cover.
llvm-svn: 345908
Instruction mapping in the outliner uses "illegal numbers" to signify that
something can't ever be part of an outlining candidate. This means that the
number is unique and can't be part of any repeated substring.
Because each of these is unique, we can use a single unique number to represent
a range of things we can't outline.
The outliner tries to leverage this using a flag which is set in an MBB when
the previous instruction we tried to map was "illegal". This patch improves
that logic to work across MBBs. As a bonus, this also simplifies the mapping
logic somewhat.
This also updates the machine-outliner-remarks test, which was impacted by the
order of Candidates on an OutlinedFunction changing. This order isn't
guaranteed, so I added a FIXME to fix that in a follow-up. The order of
Candidates on an OutlinedFunction isn't important, so this still is NFC.
llvm-svn: 345906
Summary: Different variants of idot8 codegen dag patterns are not generated by llvm-tablegen due to a huge
increase in the compile time. Support the pattern that clang FE generates after reordering the
additions in integer-dot8 source language pattern.
Author: FarhanaAleen
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D53937
llvm-svn: 345902
Summary:
This function was causing a crash when `MaxElements == 1` because
it was trying to create a single element vector type.
Reviewers: dsanders, aemerson, aditya_nandakumar
Reviewed By: dsanders
Subscribers: rovka, kristof.beyls, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D53734
llvm-svn: 345875
The test causes a crash because we were trying to extract v4f32 to v3f32, and the
narrowing factor was then 4/3 = 1 producing a bogus narrow type.
This should fix:
https://bugs.llvm.org/show_bug.cgi?id=39511
llvm-svn: 345842
Summary:
This adds dummy implementation of `EmitRawText` in `MCNullStreamer`.
This fixes the behavior of `AsmPrinter` with `MCNullStreamer` on targets
on which no integrated assembler is used. An attempt to emit inline asm
on such a target would previously lead to a crash, since `AsmPrinter` does not
check for `hasRawTextSupport` in `EmitInlineAsm` and calls `EmitRawText`
anyway if integrated assembler is disabled (the behavior has changed
in D2686).
Error message printed by MCStreamer:
> EmitRawText called on an MCStreamer that doesn't support it, something
> must not be fully mc'ized
Patch by Eugene Sharygin
Reviewers: dsanders, echristo
Reviewed By: dsanders
Subscribers: eraman, llvm-commits
Differential Revision: https://reviews.llvm.org/D53938
llvm-svn: 345841
While mutating instructions, we sign extended negative constant
operands for binary operators that can safely overflow. This was to
allow instructions, such as add nuw i8 %a, -2, to still be able to
perform a subtraction. However, the code to handle constants doesn't
take into consideration that instructions, such as sub nuw i8 -2, %a,
require the i8 -2 to be converted into i32 254.
This is a relatively simple fix, but I've taken the time to
reorganise the code a bit - mainly that instructions that can be
promoted are cached and splitting up the Mutate function.
Differential Revision: https://reviews.llvm.org/D53972
llvm-svn: 345840
When matching MipsISD::JmpLink t9, TargetExternalSymbol:i32'...',
wrong JALR16_MM is selected. This patch adds missing pattern for
JmpLink, so that JAL instruction is selected.
Differential Revision: https://reviews.llvm.org/D53366
llvm-svn: 345830
Reapplying an updated version of rL345395 (reverted in rL345451), now the issues noticed in PR39483 have been fixed.
This patch allows resolveTargetShuffleInputs to remove UNDEF inputs from cases where we have more than 2 inputs.
llvm-svn: 345824
In MipsBranchExpansion::splitMBB, upon splitting
a block with two direct branches, remove the successor
of the newly created block (which inherits successors from
the original block) which is pointed to by the last
branch in the original block only if the targets of two
branches differ.
This is to fix the failing test when ran with
-verify-machineinstrs enabled.
Differential Revision: https://reviews.llvm.org/D53756
llvm-svn: 345821
Summary:
Also reduce the test case for implicit defs and test it with all
register classes.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D53855
llvm-svn: 345794
I think this is the actual important property; the previous visibility
check was an approximation.
Differential Revision: https://reviews.llvm.org/D53852
llvm-svn: 345790
Shows up rarely for 64-bit arithmetic, more frequently for the compare
patterns added in r325323.
Differential Revision: https://reviews.llvm.org/D53848
llvm-svn: 345782
SimplifySetCC could shrink a load without checking for
profitability or legality of such shink with a target.
Added checks to prevent shrinking of aligned scalar loads
in AMDGPU below dword as scalar engine does not support it.
Differential Revision: https://reviews.llvm.org/D53846
llvm-svn: 345778
lowerRangeToAssertZExt currently relies on something like EarlyCSE having
eliminated the constant range [0,1). At -O0 this leads to an assert.
Differential Revision: https://reviews.llvm.org/D53888
llvm-svn: 345770
This feature is only relevant to shaders, and is no longer used. When disabled,
lowering of reserved registers for shaders causes a compiler crash.
Remove the feature and add a test for compilation of shaders at OptNone.
Differential Revision: https://reviews.llvm.org/D53829
llvm-svn: 345763
Before this patch DbgInfoAvailable was set to true in
DwarfDebug::beginModule() or CodeViewDebug::CodeViewDebug(). This made
MIR testing weird since passes would suddenly stop dealing with debug
info just because we stopped the pipeline before the debug printers.
This patch changes the logic to initialize DbgInfoAvailable based on the
fact that debug_compile_units exist in the llvm Module. The debug
printers may then override it with false in case of debug printing being
disabled.
Differential Revision: https://reviews.llvm.org/D53885
llvm-svn: 345740
Summary:
Instead of writing boolean values temporarily into 32-bit VGPRs
if they are involved in PHIs or are observed from outside a loop,
we use bitwise masking operations to combine lane masks in a way
that is consistent with wave control flow.
Move SIFixSGPRCopies to before this pass, since that pass
incorrectly attempts to move SGPR phis to VGPRs.
This should recover most of the code quality that was lost with
the bug fix in "AMDGPU: Remove PHI loop condition optimization".
There are still some relevant cases where code quality could be
improved, in particular:
- We often introduce redundant masks with EXEC. Ideally, we'd
have a generic computeKnownBits-like analysis to determine
whether masks are already masked by EXEC, so we can avoid this
masking both here and when lowering uniform control flow.
- The criterion we use to determine whether a def is observed
from outside a loop is conservative: it doesn't check whether
(loop) branch conditions are uniform.
Change-Id: Ibabdb373a7510e426b90deef00f5e16c5d56e64b
Reviewers: arsenm, rampitec, tpr
Subscribers: kzhuravl, jvesely, wdng, mgorny, yaxunl, dstuttard, t-tye, eraman, llvm-commits
Differential Revision: https://reviews.llvm.org/D53496
llvm-svn: 345719
Summary:
The optimization to early break out of loops if all threads are dead was
never fully implemented.
But the PHI node analyzing is actually causing a number of problems, so
remove all the extra code for it.
(This does actually regress code quality in a few places because it
ends up relying more heavily on phi's of i1, which we don't do a
great job with. However, since it fixes real bugs in the wild, we
should take this change. I have some prototype changes to improve
i1 lowering in general -- not just for control flow -- which should
help recover the code quality, I just need to make those changes
fit for general consumption. -- Nicolai)
Change-Id: I6fc6c6c8961857ac6009fcfb9f7e5e48dc23fbb1
Patch-by: Christian König <christian.koenig@amd.com>
Reviewers: arsenm, rampitec, tpr
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D53359
llvm-svn: 345718
Our a16 support was only enabled for sample/gather and buffer
load/store, but not for image load/store operations (which take an i16
as the pixel index rather than a half).
Fix our isel lowering and add test cases to prove it out.
Differential Revision: https://reviews.llvm.org/D53750
llvm-svn: 345710
Emit pseudo instructions indicating unwind codes corresponding to each
instruction inside the prologue/epilogue. These are used by the MCLayer to
populate the .xdata section.
Differential Revision: https://reviews.llvm.org/D50288
llvm-svn: 345701
The debug-use flag must be set exactly for uses on DBG_VALUEs. This is
so obvious that it can be trivially inferred while parsing. This will
reduce noise when printing while omitting an information that has little
value to the user.
The parser will keep recognizing the flag for compatibility with old
`.mir` files.
Differential Revision: https://reviews.llvm.org/D53903
llvm-svn: 345671
Summary:
Thunk functions in Windows are varag functions that call a musttail function
to pass the arguments after the fixup is done. We need to make sure that we
forward the arguments from the caller vararg to the callee vararg function.
This is the same mechanism that is used for Windows on X86.
Reviewers: ssijaric, eli.friedman, TomTan, mgrang, mstorsjo, rnk, compnerd, efriedma
Reviewed By: efriedma
Subscribers: efriedma, kristof.beyls, chrib, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D53843
llvm-svn: 345641
Summary:
Normalize the offset for endianess before checking
if the store cover the load in ForwardStoreValueToDirectLoad.
Without this we missed out on some optimizations for big
endian targets. If for example having a 4 bytes store followed
by a 1 byte load, loading the least significant byte from the
store, the STCoversLD check would fail (see @test4 in
test/CodeGen/AArch64/load-store-forwarding.ll).
This patch also fixes a problem seen in an out-of-tree target.
The target has i40 as a legal type, it is big endian,
and the StoreSize for i40 is 48 bits. So when normalizing
the offset for endianess we need to take the StoreSize into
account (assuming that padding added when storing into
a larger StoreSize always is added at the most significant
end).
Reviewers: niravd
Reviewed By: niravd
Subscribers: javed.absar, kristof.beyls, llvm-commits, uabelho
Differential Revision: https://reviews.llvm.org/D53776
llvm-svn: 345636
Prevents the post-RA scheduler from modifying the prologue sequences
emitting by frame lowering. This is roughly similar to what we do for
other targets: TargetInstrInfo::isSchedulingBoundary checks
isPosition(), which checks for CFI_INSTRUCTION.
isSEHInstruction is taken from D50288; it'll land with whatever patch
lands first.
Differential Revision: https://reviews.llvm.org/D53851
llvm-svn: 345634
Re-apply r345315 with testcase fixes.
Include all of the store's source vector operands when creating the
MachineMemOperand. Previously, we were missing the first operand,
making the store size seem smaller than it really is.
Differential Revision: https://reviews.llvm.org/D52816
llvm-svn: 345631
The CONCAT_VECTORS case was using the original mask element count to determine how to adjust the broadcast index. But if we looked through a bitcast the original mask size doesn't tell us anything about the concat_vectors.
This patch switchs to using the concat_vectors input element count directly instead.
Differential Revision: https://reviews.llvm.org/D53823
llvm-svn: 345626
The SchedModel allows the addition of ReadAdvances to express that certain
operands of the instructions are needed at a later point than the others.
RegAlloc may add pseudo operands that are not part of the instruction
descriptor, and therefore cannot have any read advance entries. This meant
that in some cases the desired read advance was nullified by such a pseudo
operand, which still had the original latency.
This patch fixes this by making sure that such pseudo operands get a zero
latency during DAG construction.
Review: Matthias Braun, Ulrich Weigand.
https://reviews.llvm.org/D49671
llvm-svn: 345606
Narrowing vector binops came up in the demanded bits discussion in D52912.
I don't think we're going to be able to do this transform in IR as a canonicalization
because of the risk of creating unsupported widths for vector ops, but we already have
a DAG TLI hook to allow what I was hoping for: isExtractSubvectorCheap(). This is
currently enabled for x86, ARM, and AArch64 (although only x86 has existing regression
test diffs).
This is artificially limited to not look through bitcasts because there are so many
test diffs already, but that's marked with a TODO and is a small follow-up.
Differential Revision: https://reviews.llvm.org/D53784
llvm-svn: 345602
Summary:
Because of the D48768, that pattern is always unfolded into pattern d,
thus we had no test coverage.
Reviewers: RKSimon, craig.topper
Reviewed By: craig.topper
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D53574
llvm-svn: 345583
Similar to FoldCONCAT_VECTORS, this patch adds FoldBUILD_VECTOR to simplify cases that can avoid the creation of the BUILD_VECTOR - if all the operands are UNDEF or if the BUILD_VECTOR simplifies to a copy.
This exposed an assumption in some AMDGPU code that getBuildVector was guaranteed to be a BUILD_VECTOR node that I've tried to handle.
Differential Revision: https://reviews.llvm.org/D53760
llvm-svn: 345578
Summary: Previously if we had a bitcast vector output type that needs promotion and a vector input type that needs widening we would just do a stack store and load to handle the conversion. We can do a little better if we can widen the bitcast to a legal vector type the same size as the widened input type. Then we can do the bitcast between this widened type and the widened input type. Afterwards we can extract_subvector back to the original output and any_extend that. Type legalization will then circle back and handle promotion of the extract_subvector and the any_extend will just be removed. This will avoid going through the stack and allows us to remove a custom version of this legalization from X86.
Reviewers: efriedma, RKSimon
Reviewed By: efriedma
Subscribers: javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D53229
llvm-svn: 345567
If a function has target features, it may contain instructions that aren't
represented in the default set of instructions. If the outliner pulls out one
of these instructions, and the function doesn't have the right attributes
attached, we'll run into an LLVM error explaining that the target doesn't
support the necessary feature for the instruction.
This makes outlined functions inherit target features from their parents.
It also updates the machine-outliner.ll test to check that we're properly
inheriting target features.
llvm-svn: 345535
- Relex hard coded registers and stack frame sizes
- Some test cleanups
- Change phi-dbg.ll to match on mir output after phi elimination instead
of going through the whole codegen pipeline.
This is in preparation for https://reviews.llvm.org/D52010
I'm committing all the test changes upfront that work before and after
independently.
llvm-svn: 345532
The machine verifier was disabled for x86 by default. There are now only
9 tests failing, compared to what previously was between 20 and 30.
This is a good opportunity to file bugs for all the remaining issues,
then explicitly disable the failing tests and enabling the machine
verifier by default.
This allows us to avoid adding new tests that break the verifier.
PR27481
llvm-svn: 345513
Add an intrinsic that takes 2 integers and perform saturation subtraction on
them.
This is a part of implementing fixed point arithmetic in clang where some of
the more complex operations will be implemented as intrinsics.
Differential Revision: https://reviews.llvm.org/D53783
llvm-svn: 345512
- Add support to generate AUTIBSP, PACIBSP, RETAB instructions for return
address signing
- The key used to sign the function is controlled by the function attribute
"sign-return-address-key"
Differential Revision: https://reviews.llvm.org/D51427
llvm-svn: 345511
This test breaks the X86 MachineVerifier. It looks like the MIR part is
completely useless.
The original author suggests that it can be removed.
Differential Revision: https://reviews.llvm.org/D53767
llvm-svn: 345501
When the floating point constants are whole numbers they have no decimal point so look like integers, but mean something very different in something like an 'and' instruction.
Ideally we would just print a decimal point and a 0, but I couldn't see how to make APFloat::toString do that.
llvm-svn: 345488
Add vector support to TargetLowering::expandFP_TO_UINT.
This exposes an issue in X86TargetLowering::LowerVSELECT which was assuming that the select mask was the same width as the LHS/RHS ops - as long as the result is a sign splat we can easily sext/trunk this.
llvm-svn: 345473
Adding the baseline tests in a preparatory NFC commit,
so that the actual commit shows the *diff*.
Yes, i'm aware that a few of these codegen-based sched tests
are testing wrong instructions, i will fix that afterwards.
For https://reviews.llvm.org/D52779
llvm-svn: 345462
Add ARM64 unwind codes to MCLayer, as well SEH directives that will be emitted
by the frame lowering patch to follow. We only emit unwind codes into object
object files for now.
Differential Revision: https://reviews.llvm.org/D50166
llvm-svn: 345450
On GreenDragon, CodeGen/X86/cpus-no-x86_64.ll was still timing out even
after breaking up the original test. I further split off the intel and
AMD cpus which hopefully resolves this.
http://green.lab.llvm.org/green/job/clang-stage2-cmake-RgSan/
llvm-svn: 345438
Summary:
The main challenge here is that X86InstrInfo::AnalyzeBranch doesn't
understand the way we're using a CALL instruction as a branch, so we
can't list the CallTarget MBB as a successor of the entry block. If we
don't list it as a successor, then the AsmPrinter doesn't print a label
for the MBB.
Fix the issue by inserting our own label at the beginning of the call
target block. We can rely on the AsmPrinter to always emit it, even
though the block appears to be unreachable, but address-taken.
Fixes PR38391.
Reviewers: thegameg, chandlerc, echristo
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D53653
llvm-svn: 345426
The "dead" markings allow existing target-independent optimizations,
like MachineSink, to trigger more frequently. The CPSR defs would have
eventually been marked dead by LiveVariables, so this only affects
optimizations before regalloc.
The ARMBaseInstrInfo.cpp change is fixing a bug which is only visible
with this change: the transform adds a use to an otherwise dead def
of CPSR. This is covered by existing regression tests.
thumb2-tbh.ll breaks for Thumb1 due to MachineLICM changing the
generated code; I'll fix it in D53452.
Differential Revision: https://reviews.llvm.org/D53453
llvm-svn: 345420
Currently, for this node:
vector int test(int a, int b, int c, int d) {
return (vector int) { a, b, c, d };
}
we get this on Power9:
mtvsrdd 34, 5, 3
mtvsrdd 35, 6, 4
vmrgow 2, 3, 2
and this on Power8:
mtvsrwz 0, 3
mtvsrwz 1, 5
mtvsrwz 2, 4
mtvsrwz 3, 6
xxmrghd 34, 1, 0
xxmrghd 35, 3, 2
vmrgow 2, 3, 2
This can be improved to this on LE Power9:
rldimi 3, 4, 32, 0
rldimi 5, 6, 32, 0
mtvsrdd 34, 5, 3
and this on LE Power8
rldimi 3, 4, 32, 0
rldimi 5, 6, 32, 0
mtvsrd 34, 3
mtvsrd 35, 5
xxpermdi 34, 35, 34, 0
This patch updates the TD pattern to generate the optimized sequence for both
Power8 and Power9 on LE and BE.
Differential Revision: https://reviews.llvm.org/D53494
llvm-svn: 345414
These promotions add additional bitcasts to the SelectionDAG that can pessimize computeKnownBits/computeNumSignBits. It also seems to interfere with broadcast formation.
This patch removes the promotion and adds isel patterns instead.
The increased table size is more than I would like, but hopefully we can find some canonicalizations or other tricks to start pruning out patterns going forward.
Differential Revision: https://reviews.llvm.org/D53268
llvm-svn: 345408
This is a narrow fix for 1 of the problems mentioned in PR27780:
https://bugs.llvm.org/show_bug.cgi?id=27780
I looked at more general solutions, but it's a mess. We canonicalize shuffle masks
based on the number of elements accessed from each operand, and that's not optional.
If you remove that, we'll crash because we fail to match isel patterns. So I'm
waiting until we're sure that we have blendvb with constant condition and then
commuting based on the load potential. Other cases like blend-with-immediate are
already handled elsewhere, so this is probably not a common problem anyway.
I didn't use "MayFoldLoad" because that checks for one-use and in these cases, we've
screwed that up by creating a temporary PSHUFB using these operands that we're counting
on to be killed later. Undoing that didn't look like a simple task because it's
intertwined with determining if we actually use both operands of the shuffle or not.a
Differential Revision: https://reviews.llvm.org/D53737
llvm-svn: 345390
While working on FileCheck producing better diagnostics in D53710, I noticed
that our test case is broken in a few different ways. The test was running, but
results were not checked as prefix CHECK-COMMON wasn't defined (which is what
FileCheck should warn about). Also, the output was different in 2 cases because
of recent changes in ARMCodeGenPrepare.
Differential Revision: https://reviews.llvm.org/D53746
llvm-svn: 345386
AMDGPU currently only supports direct calls, but at lower optimisation levels it
fails to lower statically direct calls which appear indirect due to a bitcast.
Add a pass to visit all CallSites and use CallPromotionUtils to "devirtualize"
calls.
Differential Revision: https://reviews.llvm.org/D52741
llvm-svn: 345382
.debug_loclists is the DWARF 5 version of the .debug_loc.
With that patch, it will be emitted when DWARF 5 is used.
Differential revision: https://reviews.llvm.org/D53365
llvm-svn: 345377
For both operands are bool, short, int, long, long long, add the following optimization.
1. 0-x == y --> x+y ==0
2. 0-x != y --> x+y != 0
Review: nemanjai
Differential Revision: https://reviews.llvm.org/D53360
llvm-svn: 345366
For both operands are bool, short, int, long, long long, add the following optimization test case.
1. 0-x == y --> x+y ==0
2. 0-x != y --> x+y != 0
Review: nemanjai
Differential Revision: https://reviews.llvm.org/D53358
llvm-svn: 345365
At present a v2i16 -> v2f64 convert is implemented by extracts to scalar,
scalar converts, and merge back into a vector. Use vector converts instead,
with the int data permuted into the proper position and extended if necessary.
Patch by RolandF.
Differential revision: https://reviews.llvm.org/D53346
llvm-svn: 345361
In order to print the IR slot number for the memory operand, the DAG pointer
must be passed to SDNode::dump().
The isel-debug.ll test updated to also check for the IR Value reference being
printed correctly.
Review: Ulrich Weigand
https://reviews.llvm.org/D53333
llvm-svn: 345347
Summary:
This adds support for LSDA (exception table) generation for wasm EH.
Wasm EH mostly follows the structure of Itanium-style exception tables,
with one exception: a call site table entry in wasm EH corresponds to
not a call site but a landing pad.
In wasm EH, the VM is responsible for stack unwinding. After an
exception occurs and the stack is unwound, the control flow is
transferred to wasm 'catch' instruction by the VM, after which the
personality function is called from the compiler-generated code. (Refer
to WasmEHPrepare pass for more information on this part.)
This patch:
- Changes wasm.landingpad.index intrinsic to take a token argument, to
make this 1:1 match with a catchpad instruction
- Stores landingpad index info and catch type info MachineFunction in
before instruction selection
- Lowers wasm.lsda intrinsic to an MCSymbol pointing to the start of an
exception table
- Adds WasmException class with overridden methods for table generation
- Adds support for LSDA section in Wasm object writer
Reviewers: dschuff, sbc100, rnk
Subscribers: mgorny, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D52748
llvm-svn: 345345
Handle the case where getCurrentFunction() returns nullptr by passing -1 to
printIRSlotNumber(). This will result in <badref> being printed instead of an
assertion failure.
Review: Francis Visoiu Mistrih
https://reviews.llvm.org/D53333
llvm-svn: 345342
Add LLVM intrinsics for the ARMv8.2-A FP16FML vector-form instructions. Add a
DAG pattern to define the indexed-form intrinsics in terms of the vector-form
ones, similarly to how the Dot Product intrinsics were implemented.
Based on a patch by Gao Yiling.
Differential Revision: https://reviews.llvm.org/D53632
llvm-svn: 345337
Summary:
Currently InstPrinter ignores if there are mismatches between block/loop
and end markers by skipping the case if ControlFlowStack is empty. I
guess it is better to explicitly error out in this case, because this
signals invalid input.
Reviewers: aardappel
Subscribers: dschuff, sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D53620
llvm-svn: 345333
The artificial dependencies are not real dependencies. In some cases, they
form circuits with bigger MII. However, they are used to schedule instructions
better.
Differential Revision: https://reviews.llvm.org/D53450
llvm-svn: 345319
The required-vector-width attribute was only used for backend testing and has never been generated by clang.
I believe clang is now generating min-legal-vector-width for vector uses in user code.
With this I believe passing -mprefer-vector-width=256 to clang should prevent use of zmm registers in the generated assembly unless the user used a 512-bit intrinsic in their source code.
llvm-svn: 345317
FENTRY_CALL is actually not taking any input / output operands. The
machine verifier complains now because the target description says that:
* It needs 1 unknown output
* It needs 1 or more variable inputs
llvm-svn: 345316
Include all of the store's source vector operands when creating the
MachineMemOperand. Previously, we were missing the first operand,
making the store size seem smaller than it really is.
Differential Revision: https://reviews.llvm.org/D52816
llvm-svn: 345315
KNL is based on a modified Silvermont core so I don't think these features apply. I think the LEA flag is probably also wrong, but I'm less sure as I barely understand the 3 LEA flags we have currently.
Differential Revision: https://reviews.llvm.org/D53671
llvm-svn: 345285
Summary:
Currently, Legalizer is trying to lower G_LOAD with a vector type
that has more than two elements due to the incorrect LegalityPredicate.
This patch fixes the issue by removing the multiplication by 8
as `MemDesc.Size` already contains the size in bits.
Reviewers: dsanders, aemerson
Reviewed By: dsanders
Subscribers: rovka, javed.absar, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D53679
llvm-svn: 345282
If we have a 64-bit EXT where one of the operands is a subvector of a 128-bit
vector then in some cases we can eliminate an extract_subvector by converting
to a 128-bit EXT of the 128-bit vector.
Differential Revision: https://reviews.llvm.org/D53582
llvm-svn: 345275
Currently a vector move of 0 or -1 will use different instructions depending on
the size of the vector. Using a single instruction (the 128-bit one) for both
gives more opportunity for Machine CSE to eliminate instructions.
Differential Revision: https://reviews.llvm.org/D53579
llvm-svn: 345270
The current state of the llc invocation is:
* Running all the passes from dwarfehprepare to stack coloring
(included)
* It runs it from the LLVM IR included in the file
* It *ADDS* the generated MI from ISel to the MI in the MIR file
* The machine verifier doesn't like it.
Differential Revision: https://reviews.llvm.org/D53698
llvm-svn: 345266
As suggested on D52965, this patch moves the i64 to f64 UINT_TO_FP expansion code from LegalizeDAG into TargetLowering and makes it available to LegalizeVectorOps as well.
Not only does this help perform X86 lowering as a true vectorization instead of (partially vectorized) scalar conversions, it avoids the HADDPD op from the scalar code which can be slow on most targets.
The AVX512F does have the vcvtusi2sdq scalar operation but we don't unroll to use it as it seems to only help for the v2f64 case - otherwise the unrolling cost will certainly be too high. My feeling is that we should leave it to the vectorizers - and if it generates the vector UINT_TO_FP we should use it.
Differential Revision: https://reviews.llvm.org/D53649
llvm-svn: 345256
When SimplifyCFG changes the PHI node into a select instruction, the debug line records becomes ambiguous. It causes the debugger to display unreachable source lines.
Differential Revision: https://reviews.llvm.org/D53287
llvm-svn: 345250
Summary: Fixes part of the problem reported in bug 39275.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits, alexcrichton
Differential Revision: https://reviews.llvm.org/D53542
llvm-svn: 345230
Summary:
Currently when assigning depths 'rethrow' does not take the whole
control flow stack into accounts but only considers EH pad stacks. When
assigning depth immmediates to rethrows, in normal cases it is done
correctly but when a rethrow instruction throws up to a caller, i.e., we
convert a pseudo RETHROW_TO_CALLER instruction to a rethrow, it
mistakenly compute the whole stack depth.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D53619
llvm-svn: 345223
Summary:
Changing the node type in lowering was violating assumptions made in
the DAG combiner, so don't change the node type any more. This fixes
one of the issues reported in bug 39275.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits, alexcrichton
Differential Revision: https://reviews.llvm.org/D53537
llvm-svn: 345221
Instead of using the MOVGOT64r pseudo, use the existing
MO_PIC_BASE_OFFSET support on symbol operands. Now I don't have to
create a "scratch register operand" for the pseudo to use, and the
register allocator can make better decisions.
Fixes some X86 verifier errors tracked in PR27481.
llvm-svn: 345219
Summary:
Changes all uses of minnan/maxnan to minimum/maximum
globally. These names emphasize that the semantic difference between
these operations is more than just NaN-propagation.
Reviewers: arsenm, aheejin, dschuff, javed.absar
Subscribers: jholewinski, sdardis, wdng, sbc100, jgravelle-google, jrtc27, atanasyan, llvm-commits
Differential Revision: https://reviews.llvm.org/D53112
llvm-svn: 345218
It's possible to do a tail call to a stack argument. LLVM already
calculates the right stack offset to call through.
Fixes the sibcall* and musttail* verifier failures tracked at PR27481.
llvm-svn: 345197
Add X86 SimplifyDemandedBitsForTargetNode and use it to simplify PMULDQ/PMULUDQ target nodes.
This enables us to repeatedly simplify the node's arguments after the previous approach had to be reverted due to PR39398.
Differential Revision: https://reviews.llvm.org/D53643
llvm-svn: 345182
The BKPT instruction is specified to cause a software breakpoint,
and at least on Linux results in a SIGTRAP. This makes it more
suitable for implementing debugtrap than TRAP (aka UDF #254), which
is specified to cause an undefined instruction exception and results
in a SIGILL on Linux.
Moreover, BKPT is not marked as a terminator, which is not only
consistent with the IR instruction but allows the analyzeBlock
function to correctly analyze a basic block containing the instruction,
which fixes an assertion failure in the machine block placement pass
previously triggered by the included test case.
Because BKPT is only supported starting with ARMv5T, we continue to
use UDF #254 when targeting v4T.
Differential Revision: https://reviews.llvm.org/D53614
llvm-svn: 345171
This patch brings back the MOV64r0 pseudo instruction for zeroing a 64-bit register. This replaces the SUBREG_TO_REG MOV32r0 sequence we use today. Post register allocation we will rewrite the MOV64r0 to a 32-bit xor with an implicit def of the 64-bit register similar to what we do for the various XMM/YMM/ZMM zeroing pseudos.
My main motivation is to enable the spill optimization in foldMemoryOperandImpl. As we were seeing some code that repeatedly did "xor eax, eax; store eax;" to spill several registers with a new xor for each store. With this optimization enabled we get a store of a 0 immediate instead of an xor. Though I admit the ideal solution would be one xor where there are multiple spills. I don't believe we have a test case that shows this optimization in here. I'll see if I can try to reduce one from the code were looking at.
There's definitely some other machine CSE(and maybe other passes) behavior changes exposed by this patch. So it seems like there might be some other deficiencies in SUBREG_TO_REG handling.
Differential Revision: https://reviews.llvm.org/D52757
llvm-svn: 345165
A lifetime end intrinsic between a tail call and the return should not
prevent the call from being tail call optimized.
Differential Revision: https://reviews.llvm.org/D53519
llvm-svn: 345163
A global alias may use indices which are not considered in bounds. In
such a case, accessing the base object will fail as it only peers
through inbounds accesses. This pattern is used by the swift compiler
to create references to preceeding members in the type metadata. This
would cause the code generation to fail when targeting a platform that
used ELF as the object file format. Be conservative and fail the
read-only check if we run into an alias that we cannot peer through.
llvm-svn: 345107
When implementing memset's today we often see this pattern:
$x0 = MOV 0xXYXYXYXYXYXYXYXY
store $x0, ...
$w1 = MOV 0xXYXYXYXY
store $w1, ...
We first create a 64bit constant in a 64bit register with all bytes the
same and then create a 32bit constant with all bytes the same in a 32bit
register. In many targets we could just access the lower byte of the
64bit register instead.
- Ideally this would be handled by the ConstantHoist pass but it runs
too early when memset isn't expanded yet.
- The memset expansion code already had this optimization implemented,
however SelectionDAG constantfolding would constantfold the
"trunc(bigconstnat)" pattern to "smallconstant".
- This patch makes the memset expansion mark the constant as Opaque and
stop DAGCombiner from constant folding in this situation. (Similar to
how ConstantHoisting marks things as Opaque to avoid folding
ADD/SUB/etc.)
Differential Revision: https://reviews.llvm.org/D53181
llvm-svn: 345102
Clearing LargeOffsetGEPMap at the end fixes a bug where if a large
offset GEP is in a dead basic block, we fail an assertion when trying
to delete the block due to the asserting VH in LargeOffsetGEPMap.
Differential Revision: https://reviews.llvm.org/D53464
llvm-svn: 345082
We can't add the MULDQ node back to the worklist after the demanded bits change has been committed in case the node has been removed entirely. This will have to wait until we have SimplifyDemandedBitsForTargetNode.
llvm-svn: 345070
As suggested on D53258, this patch shares common CTLZ expansion code between VectorLegalizer and SelectionDAGLegalize by putting it in TargetLowering.
Extension to D53474
llvm-svn: 345060
Add support to allow bit-casting from f128 to i128 and then
extracting 64 bits from the result.
Differential Revision: https://reviews.llvm.org/D49507
llvm-svn: 345053
This initially landed in rL345014, but was reverted in rL345017
due to sanitizer-x86_64-linux-fast buildbot failure in
check-lld (ELF/relocatable-versioned.s) test.
While i'm not yet quite sure what is the problem, one obvious
thing here is that extra truncation roundtrip.
Maybe that's it? If not, will re-revert.
Differential Revision: https://reviews.llvm.org/D53521
llvm-svn: 345027
Summary:
Continuation of D52348.
We also get the `c) x & (-1 >> (32 - y))` pattern here, because of the D48768.
I will add extra-uses into those tests and follow-up with a patch to handle those patterns too.
Reviewers: RKSimon, craig.topper
Reviewed By: craig.topper
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D53521
llvm-svn: 345014
Summary:
Due to previous work to make WebAssembly MC by default stack-only
inline assembly now "just works" (previously it didn't since it had
no way to know types of registers), so no further work required.
So far we only have tests (in inline-asm.ll) which test with
non-existing instructions, so this adds a test that roundtrips
both the inline assembly and its surrounding code thru the assembler.
Reviewers: dschuff, sunfish
Subscribers: sbc100, jgravelle-google, eraman, aheejin, llvm-commits
Differential Revision: https://reviews.llvm.org/D52914
llvm-svn: 344977
Add an intrinsic that takes 2 integers and perform unsigned saturation
addition on them.
This is a part of implementing fixed point arithmetic in clang where some of
the more complex operations will be implemented as intrinsics.
Differential Revision: https://reviews.llvm.org/D53340
llvm-svn: 344971
analyzeBranch()/insertBranch() etc. do not properly deal with an undef
flag on the eflags input and used to produce invalid MIR. I don't see
this ever affecting real world inputs (I don't think it is possible to
produce undef flags with llvm IR), so I simply changed the code to bail
out in this case.
rdar://42122367
llvm-svn: 344970
I've included a fix to DAGCombiner::ForwardStoreValueToDirectLoad that I believe will prevent the previous miscompile.
Original commit message:
Theoretically this was done to simplify the amount of isel patterns that were needed. But it also meant a substantial number of our isel patterns have to match an explicit bitcast. By making the vXi32/vXi16/vXi8 types legal for loads, DAG combiner should be able to change the load type to rem
I had to add some additional plain load instruction patterns and a few other special cases, but overall the isel table has reduced in size by ~12000 bytes. So it looks like this promotion was hurting us more than helping.
I still have one crash in vector-trunc.ll that I'm hoping @RKSimon can help with. It seems to relate to using getTargetConstantFromNode on a load that was shrunk due to an extract_subvector combine after the constant pool entry was created. So we end up decoding more mask elements than the lo
I'm hoping this patch will simplify the number of patterns needed to remove the and/or/xor promotion.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits, RKSimon
Differential Revision: https://reviews.llvm.org/D53306
llvm-svn: 344965
Recommits r342942, which was reverted in r343189, with a fix for an
issue where we would propagate unsafely if we defined only the upper
part of a register.
Original message:
Change the copy tracker to keep a single map of register units
instead of 3 maps of registers. This gives a very significant
compile time performance improvement to the pass. I measured a
30-40% decrease in time spent in MCP on x86 and AArch64 and much
more significant improvements on out of tree targets with more
registers.
llvm-svn: 344942
Introduce new versions that follow the IEEE semantics
to help with legalization that may need quieted inputs.
There are some regressions from inserting unnecessary
canonicalizes when these are matched from fast math
fcmp + select which should be fixed in a future commit.
llvm-svn: 344914
Summary:
Trivial continuation of D52304.
While this pattern is not canonical, we do select it in the BZHI case,
so this should not be any different.
Reviewers: RKSimon, craig.topper, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52348
llvm-svn: 344902
The D-Form VSX loads introduced in ISA 3.0 are not direct D-Form equivalent of
the corresponding X-Forms since they only target the Altivec registers.
Namely LXSSPX can load into any of the 64 VSX registers whereas LXSSP can only
load into the upper 32 VSX registers. Similarly with the remaining affected
instructions.
There is currently no way that I can see to trigger the bug, but as we add other
ways of exploiting these instructions, there may very well be instances that do.
This is an NFC patch in practical terms since the changes it introduces can not
be triggered without an MIR test.
Differential revision: https://reviews.llvm.org/D53323
llvm-svn: 344894
This makes fast isel treat all legal vector types the same way. Previously only vXi64 was in the fast-isel tables.
This unfortunately prevents matching of andn by fast-isel for these types since the requires SelectionDAG. But we already had this issue for vXi64. So at least we're consistent now.
Interestinly it looks like fast-isel can't handle instructions with constant vector arguments so the the not part of the andn patterns is selected with SelectionDAG. This explains why VPTERNLOG shows up in some of the tests.
This is a subset of D53268. As I make progress on that, I will try to reduce the number of lines in the tablegen files.
llvm-svn: 344884
Summary:
Theoretically this was done to simplify the amount of isel patterns that were needed. But it also meant a substantial number of our isel patterns have to match an explicit bitcast. By making the vXi32/vXi16/vXi8 types legal for loads, DAG combiner should be able to change the load type to remove the bitcast.
I had to add some additional plain load instruction patterns and a few other special cases, but overall the isel table has reduced in size by ~12000 bytes. So it looks like this promotion was hurting us more than helping.
I still have one crash in vector-trunc.ll that I'm hoping @RKSimon can help with. It seems to relate to using getTargetConstantFromNode on a load that was shrunk due to an extract_subvector combine after the constant pool entry was created. So we end up decoding more mask elements than the load size.
I'm hoping this patch will simplify the number of patterns needed to remove the and/or/xor promotion.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits, RKSimon
Differential Revision: https://reviews.llvm.org/D53306
llvm-svn: 344877
This is a late backend subset of the IR transform added with:
D52439
We can confirm that the conversion to a 'trunc' is correct by running:
$ opt -instcombine -data-layout="e"
(assuming the IR transforms are correct; change "e" to "E" for big-endian)
As discussed in PR39016:
https://bugs.llvm.org/show_bug.cgi?id=39016
...the pattern may emerge during legalization, so that's we are waiting for an
insertelement to become a scalar_to_vector in the pattern matching here.
The DAG allows for fun variations that are not possible in IR. Result types for
extracts and scalar_to_vector don't necessarily match input types, so that means
we have to be a bit more careful in the transform (see code comments).
The tests show that we don't handle cases that require a shift (as we did in the
IR version). I've left that as a potential follow-up because I'm not sure if
that's a real concern at this late stage.
Differential Revision: https://reviews.llvm.org/D53201
llvm-svn: 344872
Change of approach, it looks like it's a much better idea to deal with
the vregs that have LLTs and reg classes both properly, than trying to
avoid creating those across all GlobalISel passes and all targets.
The change mostly touches MachineRegisterInfo::constrainRegClass,
which is apparently only used by MachineCSE. The changes are NFC for
any pipeline but one that contains MachineCSE mid-GlobalISel.
NOTE on isCallerPreservedOrConstPhysReg change in MachineCSE:
There is no test covering it as the only way to insert a new pass
(MachineCSE) from a command line I know of is llc's -run-pass option,
which only works with MIR, but MIRParser freezes reserved registers upon
MachineFunctions creation, making it impossible to reproduce the state
that exposes the issue.
Reviwed By: aditya_nandakumar
Differential Revision: https://reviews.llvm.org/D53144
llvm-svn: 344822
Summary:
Add selection patterns to support one bit Sub.
Reviewers:
rampitec, arsenm
Differential Revision:
https://reviews.llvm.org/D52946
llvm-svn: 344815
https://reviews.llvm.org/D53304
Currently dead phis are not cleaned up during DCE. This patch allows
dead PHI and G_PHI insts to be deleted.
Reviewed by: dsanders
llvm-svn: 344811
Summary:
Undefined indices in shuffles can be used when not all lanes of the
output vector will be used. This happens for example in the expansion
of vector reduce operations. Regardless, undefs are legal as lane
indices in IR and should be supported.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D53057
llvm-svn: 344803
In a loop, create artificial dependences between the source of a
COPY/REG_SEQUENCE to the use in next iteration.
Eg:
SRC ----Data Dep--> COPY
COPY ---Anti Dep--> PHI (implies, to be used in next iteration)
PHI ----Data Dep--> USE
This patches creates
USE ----Artificial Dep---> SRC
This will effectively schedule the COPY late to eliminate additional copies.
Before this patch, the schedule can be
SRC, COPY, USE : The COPY is used in next iteration and it needs to be
preserved.
After this patch, the schedule can be
USE, SRC, COPY : The COPY is used in next iteration and the live interval is
reduced.
Differential Revision: https://reviews.llvm.org/D53303
llvm-svn: 344748
Allows to disable direct TLS segment access (%fs or %gs). GCC supports
a similar flag, it can be useful in some circumstances, e.g. when a thread
context block needs to be updated directly from user space. More info
and specific use cases: https://bugs.llvm.org/show_bug.cgi?id=16145
There is another revision for clang as well.
Related: D53102
All X86 CodeGen tests appear to pass:
```
[46/47] Running lit suite /SourceCache/llvm-trunk-8.0/test/CodeGen
Testing Time: 23.17s
Expected Passes : 3801
Expected Failures : 15
Unsupported Tests : 8021
```
Reviewed by: Craig Topper.
Patch by nruslan (Ruslan Nikolaev).
Differential Revision: https://reviews.llvm.org/D53103
llvm-svn: 344723
Summary:
To workaround a hardware issue in the (base + offset) calculation
when base is negative. The impact on code quality should be limited
since SILoadStoreOptimizer still runs afterwards and is able to
combine loads/stores based on known sign information.
This fixes visible corruption in Hitman on SI (easily reproducible
by running benchmark mode).
Change-Id: Ia178d207a5e2ac38ae7cd98b532ea2ae74704e5f
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99923
Reviewers: arsenm, mareko
Subscribers: jholewinski, kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D53160
llvm-svn: 344698
Summary:
Moving SMRD to VMEM in SIFixSGPRCopies is rather bad for performance if
the load is really uniform. So select the scalar load intrinsics directly
to either VMEM or SMRD buffer loads based on divergence analysis.
If an offset happens to end up in a VGPR -- either because a floating
point calculation was involved, or due to other remaining deficiencies
in SIFixSGPRCopies -- we use v_readfirstlane.
There is some unrelated churn in tests since we now select MUBUF offsets
in a unified way with non-scalar buffer loads.
Change-Id: I170e6816323beb1348677b358c9d380865cd1a19
Reviewers: arsenm, alex-t, rampitec, tpr
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D53283
llvm-svn: 344696
Previously reverted in rL343082.
Original commit message:
On failing to find sequences that can be converted into dual macs,
try to find sequential 16-bit loads that are used by muls which we
can then use smultb, smulbt, smultt with a wide load.
Differential Revision: https://reviews.llvm.org/D51983
llvm-svn: 344693
This is patch 2/2, following up on D53314, and is the functional change
to prevent fusing mul + add sequences into VFMAs.
Differential revision: https://reviews.llvm.org/D53315
llvm-svn: 344683
This is a follow up of rL342874, which stopped fusing muls and adds into VMLAs
for performance reasons on the Cortex-M4 and Cortex-M33. This is a serie of 2
patches, that is trying to achieve the same for VFMA. The second column in the
table below shows what we were generating before rL342874, the third column
what changed with rL342874, and the last column what we want to achieve with
these 2 patches:
--------------------------------------------------------
| Opt | < rL342874 | >= rL342874 | |
|------------------------------------------------------|
|-O3 | vmla | vmul | vmul |
| | | vadd | vadd |
|------------------------------------------------------|
|-Ofast | vfma | vfma | vmul |
| | | | vadd |
|------------------------------------------------------|
|-Oz | vmla | vmla | vmla |
--------------------------------------------------------
This patch 1/2, is a cleanup of the spaghetti predicate logic on the different
VMLA and VFMA codegen rules, so that we can make the final functional change in
patch 2/2. This also fixes a typo in the regression test added in rL342874.
Differential revision: https://reviews.llvm.org/D53314
llvm-svn: 344671
Without this we match the CMP+AND to a TEST and then match the SHR separately. I'm trusting analyzeCompare to remove the TEST during the peephole pass. Otherwise we need to check the flag users to see if they only use the Z flag.
This recovers a case lost by r344270.
Differential Revision: https://reviews.llvm.org/D53310
llvm-svn: 344649
Add an intrinsic that takes 2 integers and perform saturation addition on them.
This is a part of implementing fixed point arithmetic in clang where some of
the more complex operations will be implemented as intrinsics.
Differential Revision: https://reviews.llvm.org/D53053
llvm-svn: 344629
When a landing pad is calculated in a program that is compiled
for micromips, it will point to an even address. Such an error will
cause a segmentation fault, as the instructions in micromips are
aligned on odd addresses. This patch sets the last bit of the offset
where a landing pad is, to 1, which will effectively be
an odd address and point to the instruction exactly.
Differential Revision: https://reviews.llvm.org/D52985
llvm-svn: 344591
Summary:
This adds support for LSDA (exception table) generation for wasm EH.
Wasm EH mostly follows the structure of Itanium-style exception tables,
with one exception: a call site table entry in wasm EH corresponds to
not a call site but a landing pad.
In wasm EH, the VM is responsible for stack unwinding. After an
exception occurs and the stack is unwound, the control flow is
transferred to wasm 'catch' instruction by the VM, after which the
personality function is called from the compiler-generated code. (Refer
to WasmEHPrepare pass for more information on this part.)
This patch:
- Changes wasm.landingpad.index intrinsic to take a token argument, to
make this 1:1 match with a catchpad instruction
- Stores landingpad index info and catch type info MachineFunction in
before instruction selection
- Lowers wasm.lsda intrinsic to an MCSymbol pointing to the start of an
exception table
- Adds WasmException class with overridden methods for table generation
- Adds support for LSDA section in Wasm object writer
Reviewers: dschuff, sbc100, rnk
Subscribers: mgorny, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D52748
llvm-svn: 344575
AARCH64 equivalent to D53257 - uses widening pairwise adds on vXi8 CTPOP to support i16/i32/i64 vectors.
This is a blocker for generic vector CTPOP expansion (P32655) - this will remove the aarch64 diff from D53258.
Differential Revision: https://reviews.llvm.org/D53259
llvm-svn: 344554
This is intended to make the backend on par with functionality that was
added to the IR version of SimplifyDemandedVectorElts in:
rL343727
...and the original motivation is that we need to improve demanded-vector-elements
in several ways to avoid problems that would be exposed in D51553.
Differential Revision: https://reviews.llvm.org/D52912
llvm-svn: 344541
As I suggested on PR39281, this patch uses PADDL pairwise addition to widen from the vXi8 CTPOP result to the target vector type.
This is a blocker for moving more x86 code to generic vector CTPOP expansion (P32655 + D53258) - ARM's vXi64 CTPOP currently expands, which would generate a vXi64 MUL but ARM's custom lowering expands the general MUL case and vectors aren't well handled in LegalizeDAG - improving the CTPOP lowering was a lot easier than fixing the MUL lowering for this one case......
Differential Revision: https://reviews.llvm.org/D53257
llvm-svn: 344512
Summary:
TwoAddressInstruction pass typically rewrites
%1:short = foo %0.sub_lo:long
as
%1:short = COPY %0.sub_lo:long
%1:short = foo %1:short
when having tied operands.
If there are extra un-tied operands that uses the same reg and
subreg, such as the second and third inputs to fie here:
%1:short = fie %0.sub_lo:long, %0.sub_hi:long, %0.sub_lo:long
then there was a bug which replaced the register %0 also for
the un-tied operand, but without changing the subregister indices.
So we used to get:
%1:short = COPY %0.sub_lo:long
%1:short = fie %1, %1.sub_hi:short, %1.sub_lo:short
With this fix we instead get:
%1:short = COPY %0.sub_lo:long
%1:short = fie %1, %0.sub_hi:long, %1
Reviewers: arsenm, JesperAntonsson, kparzysz, MatzeB
Reviewed By: MatzeB
Subscribers: bjope, kparzysz, wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D36224
llvm-svn: 344492
Summary:
I've noticed that the bitcasts we introduce for these make computeKnownBits and computeNumSignBits not work well in LegalizeVectorOps. LegalizeVectorOps legalizes bottom up while LegalizeDAG legalizes top down. The bottom up strategy for LegalizeVectorOps means operands are legalized before their uses. So we promote and/or/xor before we legalize the operands that use them making computeKnownBits/computeNumSignBits in places like LowerTruncate suboptimal. I looked at changing LegalizeVectorOps to be top down as well, but that was more disruptive and caused some regressions. I also looked at just moving promotion of binops to LegalizeDAG, but that had a few issues one around matching AND,ANDN,OR into VSELECT because I had to create ANDN as vXi64, but the other nodes hadn't legalized yet, I didn't look too hard at fixing that.
This patch seems to produce better results overall than my other attempts. We now form broadcasts of constants better in some cases. For at least some of them the AND was being introduced in LegalizeDAG, promoted to vXi64, and the BUILD_VECTOR was also legalized there. I think we got bad ordering of that. Now the promotion is out of the legalizer so we handle this better.
In the longer term I think we really should evaluate whether we should be doing this promotion at all. It's really there to reduce isel pattern count, but I'm wondering if we'd be better served just eating the pattern cost or doing C++ based isel for vector and/or/xor in X86ISelDAGToDAG. The masked and/or/xor will definitely be difficult in patterns if a bitcast gets between the vselect and the and/or/xor node. That becomes a lot of permutations to cover.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D53107
llvm-svn: 344487
Summary: This is similar to what D52528 did for loads. It should match what generic type legalization does in 64-bit mode where it uses a v2i64 cast and an i64 store.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D53173
llvm-svn: 344470
Summary:
getShiftAmountTy for X86 returns MVT::i8. If a BSWAP or BITREVERSE is created that requires promotion and the difference between the original VT and the promoted VT is more than 255 then we won't able to create the constant.
This patch adds a check to replace the result from getShiftAmountTy to MVT::i32 if the difference won't fit. This should get legalized later when the shift is ultimately expanded since its clearly an illegal type that we're only promoting to make it a power of 2 bit width. Alternatively we could base the decision completely on the largest shift amount the promoted VT could use.
Vectors should be immune here because getShiftAmountTy always returns the incoming VT for vectors. Only the scalar shift amount can be changed by the targets.
Reviewers: eli.friedman, RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D53232
llvm-svn: 344460
Use isConstantSplat instead of ISD::isConstantSplatVector to let us us peek through to illegal types (in this case for i686 targets to recognise i64 constants)
llvm-svn: 344452
If we have better CTLZ support than CTPOP, then use cttz(x) = width - ctlz(~x & (x - 1)) - and remove the CTTZ_ZERO_UNDEF handling as it no longer gives better codegen.
Similar to rL344447, this is also closer to LegalizeDAG's approach
llvm-svn: 344448
This patch changes the vector CTTZ lowering from:
cttz(x) = ctpop((x & -x) - 1)
to:
cttz(x) = ctpop(~x & (x - 1))
Not only does this make better use of the PANDN instruction, but it also matches the LegalizeDAG method which should allow us to remove the x86 specific code at some point in the future (we need to fix some issues with the bitcasted logic ops and CTPOP lowering first).
Differential Revision: https://reviews.llvm.org/D53214
llvm-svn: 344447
Add shuffle lowering for the case where we can shuffle the lanes into place followed by an in-lane permute.
This is mainly for cases where we can have non-repeating permutes in each lane, but for now I've just enabled it for v4f64 unary shuffles to fix PR39161 - there is no test coverage for other shuffles that might benefit yet.
We now have several cross-lane shuffle lowering methods that all do something similar - I've looked at merging some of these (notably by making the repeated mask mechanism in lowerVectorShuffleByMerging128BitLanes optional), but there is a lot of assertions/assumptions in the way that makes this tricky - I ended up going for adding yet another relatively simple method instead.
Differential Revision: https://reviews.llvm.org/D53148
llvm-svn: 344446
Summary:
AArch64 can fold some shift+extend operations on the RHS operand of
comparisons, so swap the operands if that makes sense.
This provides a fix for https://bugs.llvm.org/show_bug.cgi?id=38751
Reviewers: efriedma, t.p.northover, javed.absar
Subscribers: mcrosier, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D53067
llvm-svn: 344439
SelectionDAGBuilder::visitShift will always zero-extend a shift amount when it
is promoted to the ShiftAmountTy. This results in zero-extension (masking)
which is unnecessary for RISC-V as the shift operations only read the lower 5
or 6 bits (RV32 or RV64).
I initially proposed adding a getExtendForShiftAmount hook so the shift amount
can be any-extended (D52975). @efriedma explained this was unsafe, so I have
instead eliminate the unnecessary and operations at instruction selection time
in a manner similar to X86InstrCompiler.td.
Differential Revision: https://reviews.llvm.org/D53224
llvm-svn: 344432
This is more consistent with what we usually do and matches some code X86 custom emits in some cases that I think I can cleanup.
The MIPS test change just looks to be an instruction ordering change.
llvm-svn: 344422
This saves a conversion to extracts and build_vector. We already do this when both the result and the input need to be widened to the same type.
This changed the sse-intrinsics-fast-isel test because we don't lower (insert_vector_elt (scalar_to_vector X), Y, 1) well. We turn it into (vector_shuffle (scalar_to_vector X), (scalar_to_vector Y), <0, 4, 2, 3>) losing track of the fact that the upper elts could be undef.
We should probably find a way to prevent the scalarization of the <2 x f32> load on these tests.
llvm-svn: 344404
This is the planned follow-up to D52997. Here we are reducing horizontal vector math codegen
by default. AMD Jaguar (btver2) should have no difference with this patch because it has
fast-hops. (If we want to set that bit for other CPUs, let me know.)
The code changes are small, but there are many test diffs. For files that are specifically
testing for hops, I added RUNs to distinguish fast/slow, so we can see the consequences
side-by-side. For files that are primarily concerned with codegen other than hops, I just
updated the CHECK lines to reflect the new default codegen.
To recap the recent horizontal op story:
1. Before rL343727, we were producing hops for all subtargets for a variety of patterns.
Hops were likely not optimal for all targets though.
2. The IR improvement in r343727 exposed a hole in the backend hop pattern matching, so
we reduced hop codegen for all subtargets. That was bad for Jaguar (PR39195).
3. We restored the hop codegen for all targets with rL344141. Good for Jaguar, but
probably bad for other CPUs.
4. This patch allows us to distinguish when we want to produce hops, so everyone can be
happy. I'm not sure if we have the best predicate here, but the intent is to undo the
extra hop-iness that was enabled by r344141.
Differential Revision: https://reviews.llvm.org/D53095
llvm-svn: 344361
Summary:
Reland of
- r344197 "[MC][ELF] compute entity size for explicit sections"
- r344206 "[MC][ELF] Fix section_mergeable_size.ll"
after being reverted in r344278 due to build breakages from not
specifying a target triple.
Move test from test/CodeGen/Generic/ to test/MC/ELF/.
Add explicit target triple so we don't try to run
this test on non ELF targets.
Reported: https://reviews.llvm.org/D53056#1261707
Reviewers: fhahn, rnk, espindola, NoQ
Reviewed By: fhahn, rnk
Subscribers: NoQ, MaskRay, rengolin, emaste, arichardson, llvm-commits, pirama, srhines
Differential Revision: https://reviews.llvm.org/D53146
llvm-svn: 344360
If you have the string /usr/bin, prior to this patch it would not
be quoted by our YAML serializer. But a string like C:\src would
be, due to the presence of a backslash. This makes the quoting
rules of basically every single file path different depending on
the path syntax (posix vs. Windows).
While technically not required by the YAML specification to quote
forward slashes, when the behavior of paths is inconsistent it
makes it difficult to portably write FileCheck lines that will
work with either kind of path.
Differential Revision: https://reviews.llvm.org/D53169
llvm-svn: 344359
This reverts commit b86c16ad8c97dadc1f529da72a5bb74e9eaed344.
This is being reverted because I forgot to write a useful
commit message, so I'm going to resubmit it with an actual
commit message.
llvm-svn: 344358
Generalize SelectionDAGLegalize's CTLZ expansion to handle vectors - lets VectorLegalizer::ExpandCTLZ to just pass the expansion on instead of repeating the same codegen.
llvm-svn: 344349
Pull out repeated byte sum stage for popcount of vector elements > 8bits.
This allows us to simplify the LUT/BITMATH popcnt code to always assume vXi8 vectors, and also improves avx512bitalg codegen which only has access to vpopcntb/vpopcntw.
llvm-svn: 344348
The current BitPermutationSelector generates a code to build a value by tracking two types of bits: ConstZero and Variable.
ConstZero means a bit we need to mask off and Variable is a bit we copy from an input value.
This patch add third type of bits VariableKnownToBeZero caused by AssertZext node or zero-extending load node.
VariableKnownToBeZero means a bit comes from an input value, but it is known to be already zero. So we do not need to mask them.
VariableKnownToBeZero enhances flexibility to group bits, since we can avoid redundant masking for these bits.
This patch also renames "HasZero" to "NeedMask" since now we may skip masking even when we have zeros (of type VariableKnownToBeZero).
Differential Revision: https://reviews.llvm.org/D48025
llvm-svn: 344347
Fixes PR32160 by reducing the size of PSHUFB if we only use one of the lanes.
This approach can probably be generalized to handle any target shuffle (and any subvector index) but we have no test coverage at the moment.
llvm-svn: 344336
Failure was discovered upon running
projects/compiler-rt/test/builtins/Unit/divtc3_test.c
in a stage2 compiler build.
When compiling projects/compiler-rt/lib/builtins/divtc3.c,
a call to fmaxl within the divtc3 implementation had its
return values read from registers $2 and $3 instead of $f0 and $f2.
Include fmaxl in the list of long double emulation routines
to have its return value correctly interpreted as f128.
Almost exact issue here: https://reviews.llvm.org/D17760
Differential Revision: https://reviews.llvm.org/D52649
llvm-svn: 344326
I want to add another pattern here that includes scalar_to_vector,
so this makes that patch smaller. I was hoping to remove the
hasOneUse() check because it shouldn't be necessary for common
codegen, but an AMDGPU test has a comment suggesting that the
extra check makes things better on one of those targets.
llvm-svn: 344320
On 64-bit targets the generic legalize will use an i64 load and a scalar_to_vector for us. But on 32-bit targets i64 isn't legal and the generic legalizer will end up emitting two 32-bit loads. We have DAG combines that try to put those two loads back together with pretty good success.
This patch instead uses f64 to avoid the splitting entirely. I've made it do the same for 64-bit mode for consistency and to keep the load in the fp domain.
There are a few things in here that look like regressions in 32-bit mode, but I believe they bring us closer to the 64-bit mode codegen. And that the 64-bit mode code could be better. I think those issues should be looked at separately.
Differential Revision: https://reviews.llvm.org/D52528
llvm-svn: 344291
Having a constant value operand in the compound instruction
is not always profitable. This patch improves coremark by ~4% on
Hexagon.
Differential Revision: https://reviews.llvm.org/D53152
llvm-svn: 344284
Revert r344206 "[MC][ELF] Fix section_mergeable_size.ll"
They were causing failures on too many important buildbots for too long.
Please revert eagerly if your fix takes more than a couple of hours to land!
llvm-svn: 344278
This is an alternative to D53080 since I think using a BEXTR for a shifted mask is definitely an improvement when the shl can be absorbed into addressing mode. The other cases I'm less sure about.
We already have several tricks for handling an and of a shift in address matching. This adds a new case for BEXTR.
I've moved the BEXTR matching code back to X86ISelDAGToDAG to allow it to match. I suppose alternatively we could directly emit a X86ISD::BEXTR node that isel could pattern match. But I'm trying to view BEXTR matching as an isel concern so DAG combine can see 'and' and 'shift' operations that are well understood. We did lose a couple cases from tbm_patterns.ll, but I think there are ways to recover that.
I've also put back the manual load folding code in matchBEXTRFromAnd that I removed a few months ago in r324939. This gives us some more freedom to make decisions based on the ability to fold a load. I haven't done anything with that yet.
Differential Revision: https://reviews.llvm.org/D53126
llvm-svn: 344270
Summary:
As discussed in D48491, we can't really do this in the TableGen,
since we need to produce *two* instructions. This only implements
one single pattern. The other 3 patterns will be in follow-ups.
I'm not sure yet if we want to also fuse shift into here
(i.e `(x >> start) & ...`)
Reviewers: RKSimon, craig.topper, spatel
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D52304
llvm-svn: 344224
Summary:
Although the saturating float to int instructions are already
emitted from normal IR, the fpto{s,u}i instructions produce poison
values if the argument cannot fit in the result type. These intrinsics
are therefore necessary to get guaranteed defined saturating behavior.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D53004
llvm-svn: 344204
Summary:
Global variables might declare themselves to be in explicit sections.
Calculate the entity size always to prevent assembler warnings
"entity size for SHF_MERGE not specified" when sections are to be
marked merge-able.
Fixes PR31828.
Reviewers: rnk, echristo
Reviewed By: rnk
Subscribers: llvm-commits, pirama, srhines
Differential Revision: https://reviews.llvm.org/D53056
llvm-svn: 344197
Summary:
As discussed in [[ https://bugs.llvm.org/show_bug.cgi?id=38938 | PR38938 ]],
we fail to emit `BEXTR` if the mask is shifted.
We can't deal with that in `X86DAGToDAGISel` `before the address mode for the inc is selected`,
and we can't really do it in the normal DAGCombine, because we don't have generic `ISD::BitFieldExtract` node,
and if we simply turn the shifted mask into a normal mask + shift-left, it will be folded back.
So it would seem X86ISelLowering is the place to handle this.
This patch only moves the matchBEXTRFromAnd()
from X86DAGToDAGISel to X86ISelLowering.
It does not add support for the 'shifted mask' pattern.
Reviewers: RKSimon, craig.topper, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52426
llvm-svn: 344179
Summary:
GlobalISel generates incorrect code because the legalizer artifact
combiner assumes `G_[SZ]EXT (G_IMPLICIT_DEF)` is equivalent to
`G_IMPLICIT_DEF `.
Replace `G_[SZ]EXT (G_IMPLICIT_DEF)` with 0 because the top bits
will be 0 for G_ZEXT and 0/1 for the G_SEXT.
Reviewers: aditya_nandakumar, dsanders, aemerson, javed.absar
Reviewed By: aditya_nandakumar
Subscribers: rovka, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D52996
llvm-svn: 344163
Summary:
Extend analysis forwarding loads from preceeding stores to work with
extended loads and truncated stores to the same address so long as the
load is fully subsumed by the store.
Hexagon's swp-epilog-phis.ll and swp-memrefs-epilog1.ll test are
deleted as they've no longer seem to be relevant.
Reviewers: RKSimon, rnk, kparzysz, javed.absar
Subscribers: sdardis, nemanjai, hiraditya, atanasyan, llvm-commits
Differential Revision: https://reviews.llvm.org/D49200
llvm-svn: 344142
This is intended to restore horizontal codegen to what it looked like before IR demanded elements improved in:
rL343727
As noted in PR39195:
https://bugs.llvm.org/show_bug.cgi?id=39195
...horizontal ops can be worse for performance than a shuffle+regular binop, so I've added a TODO. Ideally, we'd
solve that in a machine instruction pass, but a quicker solution will be adding a 'HasFastHorizontalOp' feature
bit to deal with it here in the DAG.
Differential Revision: https://reviews.llvm.org/D52997
llvm-svn: 344141
When SimplifyCFG changes the PHI node into a select instruction, the debug line records becomes ambiguous. It causes the debugger to display unreachable source lines.
Differential Revision: https://reviews.llvm.org/D52887
llvm-svn: 344120
An upcoming patch will change the codegen for these patterns. This test case is
added now so that the patch can show the differences in codegen.
llvm-svn: 344112
Commit r343851 changed the format of the generated instructions.
An unnecessary load has been removed. Previously, a value would be moved
from r24 into a temporary register just to be copied into r30 before the
indirect call. Now, codegen immediately loads r24 into r30, saving a
MOVW instruction.
llvm-svn: 344111
For ISD::SIGN_EXTEND_INREG operation of v2i16 and v2i8 types will cause assert because they are registered as custom operation.
So that the type legalization phase will enter the custom hook, which do not handle ISD::SIGN_EXTEND_INREG operation and fall throw into unreachable assert.
Patch By: wuzish (Zixuan Wu)
Differential Revision: https://reviews.llvm.org/D52449
llvm-svn: 344109
Summary:
Subtraction from zero and floating point negation do not have the same
semantics, so fix lowering.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D52948
llvm-svn: 344107
Summary:
Also add tests to catch crashes in passes that are not normally run in
tests.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D52959
llvm-svn: 344094
We already do the following combines:
(bitcast int (and (bitcast fp X to int), 0x7fff...) to fp) -> fabs X
(bitcast int (xor (bitcast fp X to int), 0x8000...) to fp) -> fneg X
When the target has "bit preserving fp logic". This patch just extends it
to also combine:
(bitcast int (or (bitcast fp X to int), 0x8000...) to fp) -> fneg (fabs X)
As some targets have fnabs and even those that don't can efficiently lower
both the fabs and the fneg.
Differential revision: https://reviews.llvm.org/D44548
llvm-svn: 344093
This is the PPC-specific non-controversial part of
https://reviews.llvm.org/D44548 that simply enables this combine for PPC
since PPC has these instructions.
This commit will allow the target-independent portion to be truly target
independent.
llvm-svn: 344077
This may give slightly better opportunities for DAG combine to simplify with the operations before the setcc. It also matches the type the xors will eventually be promoted to anyway so it saves a legalization step.
Almost all of the test changes are because our constant pool entry is now v2i64 instead of v4i32 on 64-bit targets. On 32-bit targets getConstant should be emitting a v4i32 build_vector and a v4i32->v2i64 bitcast.
There are a couple test cases where it appears we now combine a bitwise not with one of these xors which caused a new constant vector to be generated. This prevented a constant pool entry from being shared. But if that's an issue we're concerned about, it seems we need to address it another way that just relying a bitcast to hide it.
This came about from experiments I've been trying with pushing the promotion of and/or/xor to vXi64 later than LegalizeVectorOps where it is today. We run LegalizeVectorOps in a bottom up order. So the and/or/xor are promoted before their users are legalized. The bitcasts added for the promotion act as a barrier to computeKnownBits if we try to use it during vector legalization of a later operation. So by moving the promotion out we can hopefully get better results from computeKnownBits/computeNumSignBits like in LowerTruncate on AVX512. I've also looked at running LegalizeVectorOps in a top down order like LegalizeDAG, but thats showing some other issues.
llvm-svn: 344071
As noted in D52747, if we prefer IR to use trunc for bool vectors rather
than and+icmp, we can expose codegen shortcomings as seen here with masked store.
Replace a hard-coded PCMPGT simplification with the more general demanded bits call
to improve things.
Differential Revision: https://reviews.llvm.org/D52964
llvm-svn: 344048
There are occasionally instances where AADB rewrites registers in such a way
that a reg-reg copy becomes a self-copy. Such an instruction is obviously
redundant and can be removed. This patch does precisely that.
Note that this will not remove various nop's that we insert (which are
themselves just self-copies). The reason those are left alone is that all of
them have their own opcodes (that just encode to a self-copy).
What prompted this patch is the fact that these self-copies sometimes end up
using registers that make the instruction a priority-setting nop, thereby
having a significant effect on performance.
Differential revision: https://reviews.llvm.org/D52432
llvm-svn: 344036
As discussed on D52964, this adds 256-bit *_EXTEND_VECTOR_INREG lowering support for AVX1 targets to help improve SimplifyDemandedBits handling.
Differential Revision: https://reviews.llvm.org/D52980
llvm-svn: 344019
One case left around nonsensical operands for the KILL instruction
which the machine verifier checks for nowadays. While this should not
hurt in release builds we should fix the machine verifier errors anyway.
llvm-svn: 344008
More tests related to PR39195:
https://bugs.llvm.org/show_bug.cgi?id=39195
If we limit the horizontal codegen, it may require different
constraints for FP and integer.
llvm-svn: 343994
This patch implements a pass that optimizes condition branches on x86 by
taking advantage of the three-way conditional code generated by compare
instructions.
Currently, it tries to hoisting EQ and NE conditional branch to a dominant
conditional branch condition where the same EQ/NE conditional code is
computed. An example:
bb_0:
cmp %0, 19
jg bb_1
jmp bb_2
bb_1:
cmp %0, 40
jg bb_3
jmp bb_4
bb_4:
cmp %0, 20
je bb_5
jmp bb_6
Here we could combine the two compares in bb_0 and bb_4 and have the
following code:
bb_0:
cmp %0, 20
jg bb_1
jl bb_2
jmp bb_5
bb_1:
cmp %0, 40
jg bb_3
jmp bb_6
For the case of %0 == 20 (bb_5), we eliminate two jumps, and the control height
for bb_6 is also reduced. bb_4 is gone after the optimization.
This optimization is motivated by the branch pattern generated by the switch
lowering: we always have pivot-1 compare for the inner nodes and we do a pivot
compare again the leaf (like above pattern).
This pass currently is enabled on Intel's Sandybridge and later arches. Some
reviewers pointed out that on some arches (like AMD Jaguar), this pass may
increase branch density to the point where it hurts the performance of the
branch predictor.
Differential Revision: https://reviews.llvm.org/D46662
llvm-svn: 343993
Emit a waterfall loop in the general case for a potentially-divergent Rsrc
operand. When practical, avoid this by using Addr64 instructions.
Recommits r341413 with changes to update the MachineDominatorTree when present.
Differential Revision: https://reviews.llvm.org/D51742
llvm-svn: 343992
Some necessary yak shaving before lowering *_EXTEND_VECTOR_INREG 256-bit vectors on AVX1 targets as suggested by D52964.
Differential Revision: https://reviews.llvm.org/D52970
llvm-svn: 343991
Summary:
The ISA is really supposed to support 64-bit atomics as well,
so the data type should be an overload.
Mesa doesn't use these atomics yet, in fact I noticed this
issue while trying to use the atomics from Mesa.
Change-Id: I77f58317a085a0d3eb933cc7e99308c48a19f83e
Reviewers: tpr
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, t-tye, jfb, llvm-commits
Differential Revision: https://reviews.llvm.org/D52291
llvm-svn: 343978
This commit adds a new IR level pass to the AMDGPU backend to perform
atomic optimizations. It works by:
- Running through a function and finding atomicrmw add/sub or uses of
the atomic buffer intrinsics for add/sub.
- If all arguments except the value to be added/subtracted are uniform,
record the value to be optimized.
- Run through the atomic operations we can optimize and, depending on
whether the value is uniform/divergent use wavefront wide operations
(DPP in the divergent case) to calculate the total amount to be
atomically added/subtracted.
- Then let only a single lane of each wavefront perform the atomic
operation, reducing the total number of atomic operations in flight.
- Lastly we recombine the result from the single lane to each lane of
the wavefront, and calculate our individual lanes offset into the
final result.
Differential Revision: https://reviews.llvm.org/D51969
llvm-svn: 343973
When branch target identification is enabled, we can only do indirect
tail-calls through x16 or x17. This means that the outliner can't
transform a BLR instruction at the end of an outlined region into a BR.
Differential revision: https://reviews.llvm.org/D52869
llvm-svn: 343969
When branch target identification is enabled, all indirectly-callable
functions start with a BTI C instruction. this instruction can only be
the target of certain indirect branches (direct branches and
fall-through are not affected):
- A BLR instruction, in either a protected or unprotected page.
- A BR instruction in a protected page, using x16 or x17.
- A BR instruction in an unprotected page, using any register.
Without BTI, we can use any non call-preserved register to hold the
address for an indirect tail call. However, when BTI is enabled, then
the code being compiled might be loaded into a BTI-protected page, where
only x16 and x17 can be used for indirect tail calls.
Legacy code withiout this restriction can still indirectly tail-call
BTI-protected functions, because they will be loaded into an unprotected
page, so any register is allowed.
Differential revision: https://reviews.llvm.org/D52868
llvm-svn: 343968
The Branch Target Identification extension, introduced to AArch64 in
Armv8.5-A, adds the BTI instruction, which is used to mark valid targets
of indirect branches. When enabled, the processor will trap if an
instruction in a protected page tries to perform an indirect branch to
any instruction other than a BTI. The BTI instruction uses encodings
which were NOPs in earlier versions of the architecture, so BTI-enabled
code will still run on earlier hardware, just without the extra
protection.
There are 3 variants of the BTI instruction, which are valid targets for
different kinds or branches:
- BTI C can be targeted by call instructions, and is inteneded to be
used at function entry points. These are the BLR instruction, as well
as BR with x16 or x17. These BR instructions are allowed for use in
PLT entries, and we can also use them to allow indirect tail-calls.
- BTI J can be targeted by BR only, and is intended to be used by jump
tables.
- BTI JC acts ab both a BTI C and a BTI J instruction, and can be
targeted by any BLR or BR instruction.
Note that RET instructions are not restricted by branch target
identification, the reason for this is that return addresses can be
protected more effectively using return address signing. Direct branches
and calls are also unaffected, as it is assumed that an attacker cannot
modify executable pages (if they could, they wouldn't need to do a
ROP/JOP attack).
This patch adds a MachineFunctionPass which:
- Adds a BTI C at the start of every function which could be indirectly
called (either because it is address-taken, or externally visible so
could be address-taken in another translation unit).
- Adds a BTI J at the start of every basic block which could be
indirectly branched to. This could be either done by a jump table, or
by taking the address of the block (e.g. the using GCC label values
extension).
We only need to use BTI JC when a function is indirectly-callable, and
takes the address of the entry block. I've not been able to trigger this
from C or IR, but I've included a MIR test just in case.
Using BTI C at function entries relies on the fact that no other code in
BTI-protected pages uses indirect tail-calls, unless they use x16 or x17
to hold the address. I'll add that code-generation restriction as a
separate patch.
Differential revision: https://reviews.llvm.org/D52867
llvm-svn: 343967
Support G_UDIV/G_UREM/G_SREM. The instruction selection
code is taken from FastISel with only minor tweaks to adapt
for GlobalISel.
Differential Revision: https://reviews.llvm.org/D49781
llvm-svn: 343966
When deciding if it is safe to optimize a conditional branch to a CBZ or
CBNZ the offsets of the BasicBlocks from the start of the function are
estimated. For inline assembly the generic getInlineAsmLength() function is
used to get a worst case estimate of the inline assembly by multiplying the
number of instructions by the max instruction size of 4 bytes. This
unfortunately doesn't take into account the generation of Thumb implicit IT
instructions. In edge cases such as when all the instructions in the block
are 4-bytes in size and there is an implicit IT then the size is
underestimated. This can cause an out of range CBZ or CBNZ to be generated.
The patch takes a conservative approach and assumes that every instruction
in the inline assembly block may have an implicit IT.
Fixes pr31805
Differential Revision: https://reviews.llvm.org/D52834
llvm-svn: 343960
The MachineOutliner for AArch64 transforms indirect calls into indirect
tail calls, replacing the call with the TCRETURNri pseudo-instruction.
This pseudo lowers to a BR, but has the isCall and isReturn flags set.
The problem is that TCRETURNri takes a tcGPR64 as the register argument,
to prevent indiret tail-calls from using caller-saved registers. The
indirect calls transformed by the outliner could use caller-saved
registers. This is fine, because the outliner ensures that the register
is available at all call sites. However, this causes a verifier failure
when the register is not in tcGPR64. The fix is to add a new
pseudo-instruction like TCRETURNri, but which accepts any GPR.
Differential revision: https://reviews.llvm.org/D52829
llvm-svn: 343959
The srli test in alu8.ll was a no-op, as it shifted by 8 bits. Fix this, and
also change the immediate in alu16.ll as shifted by something other than a
poewr of 8 is more interesting.
llvm-svn: 343958
Sanjay Patel